—from Epidemiological Bulletin, Vol. 23 No. 4, Diciembre 2002—
International Classification of Diseases:
Preparation of Short Lists for Data Tabulation
Introduction
The International Classification of Diseases (ICD) has expanded notably with
each successive revision. Improvements included more detailed information in
many categories, and the creation of new categories as knowledge of diseases
advanced and new ones were discovered. One of the most important changes occurred
between the Fifth and Sixth Revisions, when the ICD changed from a “classification
of causes of deaths,” with some 200 categories, to a “classification
of diseases and causes of deaths,” with more than a thousand categories.
Thus, from Bertillon’s original 1893 Classification (the “ICD-Zero”)
with a total of 161 categories and 200 codes, there is now the ICD-10 (1), implemented
worldwide since 1994, which currently contains a total of 12,421 codes distributed
in 2,036 categories.
The greatest degree of detail allows for more in-depth analyses by diseases,
either individually or in specific groups. However, it hinders the complete
tabulation of data to obtain a panoramic view of the health situation, identify
the most relevant problems and define priorities. For this reason, the ICD itself
offers shorter lists, based on the aggregation of several categories into a
single group. Bertillon’s classification, for example, was presented in
three “nomenclatures:” a short one comprised of 44 categories, an
intermediate of 99 categories, and a most detailed of 161. The ICD-10 contains
21 Chapters, which are broken down into 261 groups containing a total of 2,036
categories. It also offers four special (short or condensed) lists for tabulating
mortality data and one for morbidity.
Basic principles
Despite the availability of many short lists, it may be necessary to construct
lists that are more specific to the situation to be analyzed. Ideally, the process
to create these lists should involve using an available list closest to the
needs, testing it with real data, and then making the necessary adjustments.
It is important to remember that the comparisons between regions or countries
are possible only if the same list has been used for all the areas under study.
As mentioned below, PAHO currently offers and develops short lists for specific
uses.
For both mortality and morbidity, the process to group diseases — that
is the list to be selected — will depend essentially on the type of analysis
intended.
Basic principles for the construction of short lists for mortality and morbidity
data tabulation include:
– The organization of the list should be based on the ICD codes; the categories
of the list should be mutually exclusive.
– The categories should be as informative as possible, avoiding residuals
usually identified by the expressions “other”, “the remainder”
or “not specified.” However, in order to guarantee the inclusion of
all cases, at least one residual category (“all the others”) is necessary.
– In mortality data tabulation, the “ill-defined” causes (symptoms,
signs and abnormal clinical and laboratory findings, Chapter XVIII of the ICD-10)
should be shown separately, not as a category of the list. In morbidity analyses,
it may be necessary to present symptoms and signs as one or more categories
of the list.
– It is not necessary to organize the categories of the list as in the
chapters of the ICD. Indeed, in order to do this, several residual categories
would be needed to complete the chapters, thereby increasing the proportion
of cases in little-informative categories.
– The categories of the list can correspond to single codes of the ICD
(three-character categories, but not four-character subcategories), and codes
of different chapters or entire chapters of the ICD, in accordance with the
needs.
– The preparation of a list should be based on the current Revision of
the ICD (ICD-10). The rationale is that the list should be oriented to current
situations and used for many years. The preparation of an equivalent list for
the ICD-9 or previous revisions should be, accordingly, a secondary objective.
– The number of categories in a short list should be sufficiently broad
to meet the requirements above, but not excessive to the point of hindering
its complete presentation. The majority of the short lists used in mortality
and morbidity have between 30 and 150 categories.
– The morbidity short lists should usually be different from the ones used
for mortality. One reason is that many codes in the ICD cannot be used as an
underlying cause of death but can be used in morbidity.
Short lists for Mortality
If the objective is, for example, to obtain a panoramic view of the causes
of death of a country or region as a starting point for an analysis, a short
list such as the PAHO-6/67 list may be used. The 6/67 list has six broad groups
of causes that are divided into 67 detailed groups (see Table 1).(2)
For a more in-depth analysis, it is usually necessary to use a more specific
list. For an analysis of mortality patterns in specific population groups, the
lists should focus on the most common health problems in this group. If the
objective is, for example, to analyze infant mortality, the list should detail
the most common causes of death during the first year of life, such as infectious
and parasitic diseases, malnutrition, congenital malformations and conditions
originating in the perinatal period.
For the study of mortality in school-age children or adolescents, it is important
that the list used emphasize external causes, among others, since they usually
represent the majority of deaths. In studying mortality trends in adults, AIDS,
diabetes, neoplasms and diseases of the circulatory system are worth pointing
out, in addition to external causes. In the more advanced ages, chronic degenerative
processes, including neoplasms and cardiovascular, endocrine and metabolic diseases,
should be included.
Another important axis for grouping causes of death is using preventability
criteria, developed by Taucher and oriented toward the definition of priorities
and the evaluation of health measures and programs.(3) This type of list, which
can also be used for morbidity, has a structure similar to that of PAHO’s
6/67 list, meaning that the large groups would include diseases preventable
with the same type of measures. For example:
– Deaths avoidable by vaccination (e.g. measles or tetanus)
– Deaths avoidable by early diagnosis and timely and adequate treatment
(e.g. tuberculosis, syphilis or causes of maternal mortality)
– Deaths avoidable by application of hygienic measures, environmental sanitation,
and health education (e.g. intestinal infectious diseases, intoxication due
to air pollution)
– Deaths avoidable by application of a combination of measures (measures
included in more than one group)
– Deaths difficult to avoid with current knowledge and technological development
– Remaining deaths (causes not identified in any of the previous groups)
The preventability criteria of causes of death can vary depending on the historic
moment, the availability of technologies or resources, the experience of a given
country or region, or also when comparing with specific paradigms. An interesting
discussion on this topic appeared in the article “Avoidable Mortality:
Indicator or Target? Application in Developing Countries” published in
the Epidemiological Bulletin in 1990 [Click here
to access the article in pdf format (1 MB)].(4)
One frequently used way to tabulate causes of death is ranking the leading causes,
in order to identify priority problems and define policies and health programs.
In addition to the aforementioned basic principles, it is recommended that a
short list for that purpose have the following characteristics:
– The axis for grouping causes should have an epidemiological basis, associated
with the idea of control measures.
– Residual categories should be avoided, preferably using only one for
“all remaining causes.” Ideally, that residual category should not
contain more than 10% of the total. Furthermore, that category and the “ill-defined”
categories should not be included among the principal causes, but presented
separately.
– A balance should be sought between grouping and disaggregation, so that
the five leading causes of death in the general population may represent some
40-50% of the deaths and the ten or fifteen first causes, around two thirds.
Overloaded and frequently heterogeneous categories, such as “heart diseases”
or “malignant neoplasms”, should be avoided. Since the purpose is
to show the leading causes, diseases of low frequency such as “rare events”,
for example rabies, poliomyelitis or yellow fever, should also be avoided.
– In order to facilitate the tabulation and analysis, the list should have
a single hierarchical level. The breakdown of the categories for more elaborate
analyses can be done with other types of tabulations.
– The list should be based on the ICD-10 and on the current situation,
since it is used to support current analyses, definition of priorities, health
programs and policies. Several short lists are included in Volume 1 of the ICD-10.
The study of trends or the comparison with past situations may require another
type of tabulation.
– A preliminary version of the list should be tested with the most recent
real mortality data. The ideal is to tabulate data of areas with different levels
of health and/or age distribution, which makes it possible to test the capacity
for information and the discriminating power of the list. Furthermore, the ordering
of the data with different types of indicators can also facilitate the evaluation
of the list, for example ordering by frequency of deaths and also by Years of
Potential Life Lost (YPLL).(5)
A list that meets such criteria will probably have a total of between 40 and
80 categories. PAHO has prepared a specific list for ordering leading causes
of death. It is currently under revision and is being tested with data from
different countries. Once this process is completed, the list will be disseminated
and its use recommended.
Short lists for morbidity
As mentioned before, there usually is a need for specific lists for morbidity.
That is due mainly to the fact that the probability of death varies widely depending
on the type of disease. Some diseases present a high incidence but rarely cause
death. An example of difference between mortality and morbidity lists is that
of normal delivery. Indeed normal deliveries, which cannot be a cause of mortality,
appear in the lists of principal diagnosis for hospital discharge, which are
often used as morbidity lists.
As discussed above, the type of short list to be used is determined by the type
of analysis to be done. Lists for morbidity require the definition of an additional
element: the type of data to be used (hospital discharges, outpatient consultation,
primary care or others). For outpatient consultations and primary care, the
data are not always coded, and when they are, the ICD is not always directly
used. Various countries have developed their own lists for outpatient consultations,
usually derived from the ICD or primary care classifications, but adapted to
their specific needs. Furthermore, the possibility of obtaining clear and defined
diagnoses is smaller than in cases of hospitalization. For that reason, the
preparation of short lists for hospital discharges only is discussed below.
Contrasting with what occurs with mortality, the use of morbidity data follows
different criteria and purposes in different countries and even within countries.
The most frequent use of morbidity data is for payment of medical care, especially
hospital care, which represents a distortion of the use of data, in view of
the fact that the cost is related to the procedure and not to the diagnosis
of the disease.
The creation of Diagnosis-Related Groups (DRG) (6) was one of the solutions
for that problem, which at the same time provides a short list for tabulation.
This system classifies patients in groups that have similar characteristics,
related to the procedures and interventions they need as well as age group and
occurrence of complications. The system was originally designed in order to
improve the internal administration of the hospitals. Later, it became a mechanism
for payment of care. As a result of the reduction in hospital costs generated
by the use of the DRG in the United States, other countries are now using them
or planning to use them. However, the use of DRG requires specific studies in
each country, in order to make the necessary adjustments using the countries’
own criteria in accordance with local and national characteristics. The construction
of DRG requires a relatively large quantity of good quality data and the availability
of experts in classification, statistics, and computation. In the United States
for example, the list that was used originally presented 470 DRG. Examples of
DRG include “vaginal delivery without complicating diagnoses”, “Esophagitis,
gastroenteritis, and miscellaneous digestive disease age 70 and over and/or
substantial comorbidity and/or complication”, “Circulatory disorders
with acute myocardial infarction without cardiovascular complications, discharged
alive”, “Bronchitis and asthma age 18-69 without substantial comorbidity
and/or complication”, or “Medical back problems”.
Even if the coding of causes of death is made in the same way throughout the
world, using rules of selection for the underlying cause, the same does not
hold true for the selection of a corresponding unique diagnosis for an episode
of hospitalization or outpatient consultation. Even though there exist rules
defined for selecting a unique diagnosis for an episode of hospitalization since
the ICD-9, the majority of countries that code hospital morbidity have made
modifications and adaptations to the rules, in accordance with their interests
and needs.
The concept of underlying cause of death is clear and known: it is “the
disease or injury which initiated the chain of morbid events leading directly
to death, or the circumstances of the accident or violence which produced the
fatal injury.”1 The same does not occur with the concept of “main
condition” of a hospital discharge. The definition from the ICD-10 is:
“the condition diagnosed at the end of the episode of health care, primarily
responsible for the patient’s need for treatment or investigation. If there
is more than one such condition, the one held responsible for the greatest use
of resources should be selected.”(1)
As can be noted, the definition is neither completely clear, nor will it yield
the most useful selected diagnosis, in particular for health situation analysis,
as for example in the case of mortality data. There are at least three different
ways to select a unique diagnosis for hospital discharge, each related to a
different type of analysis:
– Reason for hospitalization: the disorder established, after study,
as primarily responsible for admission of the patient to the hospital
– Main condition: the disorder treated during the hospitalization,
considered as the most important in terms of clinical significance and resources
used
– Underlying cause of hospitalization: the underlying cause of the
disorder that caused the hospitalization of the patient.
The reason for hospitalization is related to the need for immediate care
of the patient and to the availability of human and technological resources,
while the main condition is strongly related to costs, procedures and financing
of the health system. In addition, the underlying cause is much more related
to health situation analysis, the main health problems, prevention measures,
health policies and programs.
The three types of unique diagnosis can lead to the same disease or disorder,
or to more than one problem, disorder, condition, disease, or pathology, depending
on each situation. Two examples are presented below:
1. A patient is discharged from the hospital, having been treated for an uncomplicated
acute appendicitis. The selected diagnosis will be the same with any of the
aforementioned criteria.
2. A patient of 70 years of age is admitted for a rhinoplasty in order to correct
sequelae of lacerations which occurred in a car accident two years earlier.
Before being discharged, the patient falls from the bed, fracturing her femur.
After 6 days of treatment of the fracture, she suffers an embolism and dies
12 hours later. In that case, the unique diagnosis differs depending on the
criterion used. The main condition to be selected could be the fracture of the
femur complicated by an embolism. The reason for hospitalization was a sequelae
of an injury to the nose and the underlying cause of hospitalization was a transport
accident. Finally, a fourth unique diagnosis could be selected for the underlying
cause of the death: fall from bed (external cause).
As can be noted, the type of short list to be used requires that the type of
analysis be defined but also, in the case of morbidity, that the criteria for
coding be clearly defined. Once defined, the preparation of a short list for
hospital discharges should follow the same general principles as those discussed
for the preparation of mortality lists. However, if the type of analysis to
be carried out is not a “health situation analysis”, as is done in
mortality, at least two important differences should be considered:
– If the underlying cause of hospitalization is not used, codes
corresponding to the nature of the injuries for the selected unique condition
should be used, and not those of the external cause of the injuries.
– (“Ill-defined”) signs and symptoms can be part of the list
and tabulated together with the other categories.
Conclusions
The results and conclusions of health analyses using mortality or morbidity
data can be influenced in many different ways. Some factors that come into play
are related to coverage and quality of the data including, among others, the
precision and adequacy of medical information on diagnoses, types of variables
used, coding, consistency and adequacy of the data, and the correct use of statistical
standards, for example the definitions of live birth and stillbirth.
However, one of the factors that has the most impact in the analyses is the
way data is organized and tabulated, depending on the type of data (mortality
or morbidity) and the type of analysis required. As a result, the choice of
an adequate list (or the construction of a specific one) is fundamental for
organizing the data. It allows for relevant health problems to appear in their
adequate dimension, avoiding erroneous or biased conclusions that may damage
evaluations and decisions and hinder the comparability between areas or time
periods.
| 0.00 Symptoms, signs and ill-defined conditions (R00-R99) 1.00 Communicable diseases (A00-B99, G00-G03, J00-J22) 1.01 Intestinal infectious diseases (A00-A09) 1.02 Tuberculosis (A15-A19) 1.03 Certain vector-borne diseases and rabies (A20, A44, A75-A79, A82-A84, A85.2, A90-A98, B50-57) 1.04 Certain diseases preventable by immunization (A33-A37, A80, B05, B06, B16, B17.0, B18.0-B18.1, B26) 1.05 Meningitis (A39, A87, G00-G03) 1.06 Septicemia, except neonatal (A40-A41) 1.07 HIV disease (AIDS) (B20-B24) 1.08 Acute respiratory infections (J00-J22) 1.09 Other infectious and parasitic diseases (remainder of A00-B99, i.e. A21-A32, A38, A42-A43, A46-A74, A81, A85.0-A85.1,A85.8, A86, A88-A89, A99-B04, B07-B15, B17.1-B17.8, B18.2-B19.9, B25, B27-B49, B58-B99) 2.00 Neoplasms (C00-D48) 2.01 Malignant neoplasm of stomach (C16) 2.02 Malignant neoplasm of colon and rectosigmoid junction (C18-C19) 2.03 Malignant neoplasm of digestive organs and peritoneum, except stomach and colon (C15, C17, C20-C26, C48) 2.04 Malignant neoplasm of trachea, bronchus and lung (C33-C34) 2.05 Malignant neoplasm of respiratory and intrathoracic organs, except trachea, bronchus and lung (C30-C32, C37-C39) 2.06 Malignant neoplasm of female breast (C50 in women) 2.07 Malignant neoplasm of cervix uteri (C53) 2.08 Malignant neoplasm of corpus uteri(C54) 2.09 Malignant neoplasm of uterus, part unspecified (C55) 2.10 Malignant neoplasm of prostate (C61) 2.11 Malignant neoplasm of other genitourinary organs (C51-C52, C56-C57, C60, C62-C68) 2.12 Leukemia (C91-C95) 2.13 Malignant neoplasm of lymphoid, other hematopoietic and related tissue (C81-C90, C96) 2.14 Malignant neoplasm of other and unspecified sites (remainder of C00-C97, i.e. C00-C14, C40-C47, C49, C50 in men, C58, C69-C80, C97) 2.15 Carcinoma in situ, benign neoplasms and neoplasms of uncertain or unknown behavior (D00-D48) 3.00 Diseases of the circulatory system (I00-I99) 3.01 Acute rheumatic fever and chronic rheumatic heart diseases (I00-I09) 3.02 Hypertensive diseases (I10-I15) 3.03 Ischemic heart diseases (I20-I25) 3.04 Pulmonary heart disease, diseases of pulmonary circulation and other forms of heart disease (I26-I45, I47-I49, I51) 3.05 Cardiac arrest (I46) 3.06 Heart failure (I50) 3.07 Cerebrovascular diseases (I60-I69) 3.08 Atherosclerosis (I70) 3.09 All other diseases of the circulatory system (I71-I99) 4.00 Certain conditions originating in the perinatal period (P00- P96) 4.01 Fetus and newborn affected by certain maternal conditions (P00, P04) 4.02 Fetus and newborn affected by obstetric complications, birth trauma (P01-P03, P10-P15) 4.03 Slow fetal growth, fetal malnutrition, short gestation, low birth weight (P05, P07) 4.04 Respiratory disorders specific to the perinatal period (P20-P28) 4.05 Bacterial sepsis of newborn (P36) 4.06 Remainder of certain conditions originating in the perinatal period (rest of P00-P96, i.e. P08, P29, P35, P37-P96) 5.00 External causes (V01-Y89) 5.01 Land transport accidents (V01-V89) 5.02 Other and unspecified transport accidents (V90-V99) 5.03 Falls (W00-W19) 5.04 Accidents caused by firearm discharge (W32-W34) 5.05 Accidental drowning and submersion (W65-W74) 5.06 Accidental threats to breathing (W75-W84) 5.07 Exposure to electric current (W85-W87) 5.08 Exposure to smoke, fire and flames (X00-X09) 5.09 Accidental poisoning by and exposure to noxious substances (X40-X49) 5.10 All other accidents (W20-W31, W35-W64, W88-W99, X10-X39, X50-X59, Y40-Y84) 5.11 Intentional self-harm (suicide) (X60-X84) 5.12 Assault (homicide) (X85-Y09) 5.13 Event of undetermined intent (Y10-Y34) 5.14 All other external causes (Y35-Y36, Y85-Y89) 6.00 All other diseases (D50-D89, E00-E90, F00-F99, G04-G98, H00-H59, H60-H95, J30-J98, K00-K93, L00-L99, M00-M99, N00-N99, O00-O99, Q00-Q99) 6.01 Diabetes mellitus (E10-E14) 6.02 Nutritional deficiencies and nutritional anemia (E40-E64, D50-D53) 6.03 Mental and behavioral disorders (F00-F99) 6.04 Diseases of the nervous system, except meningitis (G04-G99) 6.05 Chronic lower respiratory diseases (J40-J47) 6.06 Remainder of diseases of the respiratory system (rest of J00-J99, i.e. J30-J39, J60-J98) 6.07 Appendicitis, hernia of abdominal cavity and intestinal obstruction (K35-K46, K56) 6.08 Cirrhosis and certain other chronic diseases of liver (K70, K73, K74, K76) 6.09 All other diseases of the digestive system (rest of K00-K93, i.e. K00-K31, K50-K55, K57-K66, K71, K72, K75, K80-K93) 6.10 Diseases of the urinary system (N00-N39) 6.11 Hyperplasia of prostate (N40) 6.12 Pregnancy, childbirth and the puerperium (O00-O99) 6.13 Congenital malformations, deformations and chromosomal abnormalities (Q00-Q99) 6.14 Remainder of all other diseases (rest of A00-Q99, i.e. D55-D89, E00-E07, E15-E34, E65-E90, H00-H59, H60-H95, L00-L99, M00-M99, N41-N99) |
References:
(1) World Health Organization. Statistical Classification of Diseases
and Related Health Problems. - 10th Revision. 3v. Geneva, Switzerland: WHO;
1994.
(2) Pan American Health Organization. New 6/67 list for tabulation of ICD-10
mortality data. Epidemiological
Bulletin 1999;
20(3):4-9.
(3) Taucher E. Mortalidad desde 1955 a 1975. Tendencias y causas. CELADE, Serie
A. Chile: CELADE; 1978.
(4) Pan American Health Organization. Avoidable Mortality: Indicator or Target?
Application in Developing Countries. Epidemiological
Bulletin 1990;
11(1):1-9.
(5) Organización Panamericana de la Salud. Salud en las Américas,
Edición 1998. Nota Técnica 11. Washington, DC: PAHO; 1998:46.
(6) Kimberly JR, de Pouvourville G. The Migration of Managerial Innovation:
Diagnosis-related Groups and Health Care Administration in Western Europe. San
Francisco: Jossey-Bass Publishers, 1993:10.
Source: Prepared by Dr. Roberto Becker of PAHO’s Special
Program for Health Analysis (SHA).
Return to Index
Epidemiological Bulletin, Vol. 23 No. 4, Diciembre
2002