In PAHO HQ All PAHO PAHO + WHO

Epidemiological Bulletin
      Vol. 18, No. 1
 March 1997  


IMPLEMENTATION OF THE INTERNATIONAL STATISTICAL

CLASSIFICATION OF DISEASES AND RELATED HEALTH PROBLEMS,
TENTH REVISION (ICD-10)

For the Pan American Health Organization, implementation of the ICD-10 in the countries of the Region of the Americas is considered an important part of a larger process entailing the review and improvement of vital statistics--a process that attempts to take advantage of the training and adjustment needs arising from the introduction of the new Revision, which has served as a catalyst.

Specific aspects concerning the implementation of the ICD-10 are presented below, in addition to broader topics linked with vital statistics.

Return to top

Training of Encoders

In 1995 subregional workshops were held in Jamaica, Mexico, Paraguay, Trinidad and Tobago, and Venezuela. The purpose was to provide training in ICD-10 for experienced encoders familiar with earlier revisions to enable them to serve as trainers in their respective countries. Delegates from Argentina, Bolivia, Chile, Colombia, Costa Rica, Cuba, Dominican Republic, Ecuador, El Salvador, Guatemala, Honduras, Mexico, Nicaragua, Panama, Paraguay, Peru, Uruguay, and Venezuela participated in these workshops. Training workshops were also held in Brazil, with the participation of encoders from all states in that country.

The workshops were conducted by the three WHO Collaborating Centers for Classification of Diseases in the Region: the Venezuelan Center for Classification of Diseases (CEVECE), the Brazilian Center for Classification of Diseases (CBCD) and the Center for the Classification of Diseases for North American (NAC-NCHS). The subregional workshops received support from PAHO.

Through the PAHO/WHO Representative Offices, the following countries of the Region have announced the year in which they will begin encoding mortality data using the ICD-10:

There has been no official confirmation from the remaining countries.

National workshops were held in most countries in 1996, some of them offered by CEVECE, with PAHO support: i.e., Nicaragua (August) and Ecuador (October). In Peru, the workshop was held in September with PAHO support. Other countries are offering workshops conducted by encoders trained in the subregional workshops. Workshops are scheduled for early 1997 in Panama and Colombia.

A workshop has also been held in the city of O Grove in Galicia, Spain, presented by the Director of CEVECE and the PAHO Regional Adviser on the ICD, with the participation of encoders from all parts of Spain. The Spanish government reports that a decision has been made to implement the ICD-10 in 1998 for both mortality and morbidity data.

Return to top

Automated Selection of the Underlying Cause of Death

In the 1960s, the NCHS (U.S. National Center for Health Statistics) developed a system for automated selection of the underlying cause of death, called ACME (Automated Classification of Medical Entities). This system, which was designed to operate on mainframe computers, has been used in many countries with the ICD-8 and the ICD-9, with some local adaptations.

In addition to the advantage of providing standardized criteria for applying the selection rules and thereby increasing the comparability of data, automated selection of the underlying cause of death, either as a function of the processing system for mortality data or as an independent program, facilitates the study of multiple causality, significantly enhancing the usefulness of the data.

The result is a change in the typical profile of the encoder, who assumes a simpler, more rapid role in encoding per se. Because of his knowledge and experience, the encoder can be used to enter data, check its consistency, and validate and analyze data.

In recent years, the United States and other countries have been developing instruments to automate the encoding process, even using systems to introduce diagnostic terms directly without first encoding them using the ICD system.

One of the national versions for automated selection of the underlying cause of death, adapted to Latin America, is the "SCB" (Selection of Underlying Cause), developed in Brazil (CBCD/DATASUS), for the Ninth Revision, which is available in Portuguese and Spanish.

However, given the widespread use of the ICD-10 in the Hemisphere programmed for 1996 through 1998, broad implementation of a version for automated selection of the underlying cause of death for ICD-9 becomes unnecessary.

Furthermore, because the changes from the ICD-9 to the ICD-10 are very extensive, involving not only expanded codes but modifications in the selection rules and in the interpretation of acceptable sequences of causes as well, rapid adaptation of the versions for automated coding of the underlying cause of death from the Ninth to the Tenth Revision is impossible. It is anticipated that the so-called "decision tables" (i.e., codes, rules, and interpretations) for the ICD-10 will be available by the second half of 1997.

Return to top

Condensed Lists of Causes of Death based on the ICD-10

Although the ICD-10 provides some "special lists" for tabulating mortality and morbidity data, none of them is comparable to the "PAHO-6/61" list, prepared for the Ninth Revision, whose purpose is to permit an overview of the mortality profile, based on two levels of aggregation (six large groups of causes, with internal subdivisions).

The data must be tabulated for each list furnished by the ICD-10 in order to study its suitability for different approaches to the analysis of mortality. A list comparable to that of the 6/61 list is also being drawn up for the ICD-10, with a view to formulating a "basic" list that can maintain the sequence of the historical series, as published in Health Statistics from the Americas.

In this respect, the Program on Health Situation Analysis (SHA) has prepared the "6/65-ICD-10 (PROVISIONAL)" list, equivalent to the "6/61-ICD-9" list, which will be distributed for use and evaluation to the countries that already have data encoded with the Tenth Revision.

Return to top

Improving Vital Statistics

As mentioned earlier, the implementation of the ICD-10 represents an opportunity to develop a broad process for reviewing and improving health statistics, especially those based on mortality data. In this respect, the key points are as follows:

a) Revision of Death Certificate Models

As the primary source of information on mortality, the death certificate model should be as suited as possible to the two basic functions that it must fulfill: legal/civil and statistics/health.

The Revision includes a proposal to unify the models (most countries use different models for fetal and non-fetal deaths), to check the variables and the concepts behind them, as well as their purpose, formulation, and adaptation to the objectives of the ICD-10--that is, to include a question about the existence of a pregnancy in the case of female deaths and a fourth line in the section on the medical diagnosis. Models from several countries are being utilized to stimulate the discussion. The Program on Health Situation Analysis has a "collection" of certificates with the corresponding legislation from the countries of the Region.

b) Inssuance/Revision of Certificates of Live Birth

Only a few countries have information systems on births that are similar to those used for deaths, with a birth certificate. In this area, models from several countries are being used to contribute to the discussion on issuing and revising certificates of live birth.

c) Review of the Legal Framework of the Civil Registry

Discussions are being promoted on the legal framework of the civil registry, a key component in information systems on vital statistics. In many countries, the legislation is too specific; this can make it difficult to introduce changes in the model certificates or forms, in the institutions involved, or in the flow of information.

Furthermore, the legislation often barely makes mention of "compulsory" variables, providing instead details on each variable, which also makes refinements or updating difficult.

The Review always looks for ways to facilitate civil registry, formulating suggestions to be presented to the competent agencies. Certificates from different countries are also being used as support.

d) Review of Information Flows/Decentralization

The health sector reform process under way in virtually all the countries, the decentralization process, and current technological developments imply the need to revise the concepts that underlie the information systems, in terms of information flows and the role of each level in generating, transmitting, processing, and using data.

It is very important to promote the use of data at all levels: local, subnational, and national. However, encoding of the underlying causes should not be decentralized at all local levels, especially if it is done manually, as is the case in practically all the countries in the Region. Encoding can be done at the intermediate level (state, province, department, region). The use of data at the local level does not require selection of the underlying cause, because it is more immediate and determines activities in epidemiological surveillance and disease control, and because there is less data, allowing for individual analysis of each death, with all its diagnoses. It is also very expensive and difficult to keep encoders adequately trained and supervised at every local level.

e) Reformulating the Data Processing Systems

Implementation of the ICD-10 represents a timely opportunity for revising data processing systems, not only because of the switch from the Ninth to the Tenth Revision but also because of the changes in death certificate models and the modernization of technology.

As support for the automated processes to ensure data consistency, the PAHO Program on Health Situation Analysis has created a DBF file containing all the ICD-10 categories with an abbreviated description of up to 40 characters, an indication of the number of digits in each description, and comments on how each could be used as the underlying cause of death. Another file contains all the categories (3 digits) and subcategories (4 digits) with the constraints on their use (impossible, code not used, or unlikely) as the underlying cause or correlated with sex and/or age.

These files are being given to the countries, and discussions are being held on their utilization and adaptation to each country's needs and on the consistency of the data in the data processing systems. The files are also useful tools for reviewing the manual encoding of the underlying cause.

Other aspects of the data processing systems that have been revised are the basic definitions of outputs (tables) and the decentralization process.

f) Studies on the Coverage and Quality of Data

During the missions to the countries, the need to promote specific studies to evaluate the coverage and quality of the data is discussed, in keeping with each country's individual situation. What is usually available is a general estimate of underreporting for the country. However, the underreporting of mortality is almost never the same for all ages and tends to be higher for both the very young and the very old; nor is it homogeneous from region to region within a single country.

It would be extremely valuable for the data analysis to have parameters for adjusting/projecting data and indicators, based on knowledge of the specific constraints in terms of quality and coverage. In addition, the results would make it possible to plan specific activities or programs in the countries to overcome such constraints.

Another aspect discussed is the use of nonmedical information on causes of nonviolent deaths. In some countries of the Region, such information is encoded without distinguishing it from the medical information.

The recommendation is that medical information on the cause(s) of death be tabulated separately from nonmedical information, because the nature and characteristics of the information are different. As a rule, diagnostic skills are far less developed among nonmedical personnel, who almost always limit themselves to the most obvious or terminal manifestations of the disease and not the underlying cause. This is not true for external causes, where the underlying cause is not the type of injury that caused the death, but the circumstances that determined it--i.e., the type of violence (hit by a car, murdered with a gun, drowned in a river, etc.). In such cases, a physician is not needed for the specific information.

The most appropriate way to handle (nonviolent) causes of death in the general tables (national and international) is to consider nonmedical information on diagnoses an "unknown cause". Such information is significant at the local level. However, it is not generally comparable at the national level and still less so at the international level.

g) Analyzing Data and Disseminating Information

In most countries, the basic (if not the only) form in which mortality data is available and disseminated is the publication of annual mortality statistics. Discussion of these publications, revolves around the characteristics of the data, how tables and other forms of presentation are prepared, the level of aggregation and disaggregation, the types of data and/or appropriate indicators, etc.

The discussions also address the ways in which users access data--for example, electronic media, "on line" consultations, information dissemination through networks like the Internet, and special tables for users. Emphasis is also placed on the need to provide, together with the data itself, information on the characteristics, coverage, and possible limitations of the data.

Within the time available, a brief analysis is made of the country's mortality profile, with discussion of the uses of the data, ways to adapt it, projections, and estimates of any constraints or deficiencies in quality or coverage.

Source: Division of Health and Human Development, Health Situation Analysis Program. SHA, PAHO.

 

Return to issue index

Index of Online Bulletins