Chronic Disease Reports in the Morbidity & Mortality Weekly Report (MMWR)
INTRODUCTION In 1986, 1.58 million people in the United States died from six major chronic diseases: cardiovascular and cerebrovascular diseases, chronic obstructive pulmonary disease, chronic liver disease and cirrhosis, malignant neoplasms, and diabetes. These deaths accounted for 75% of all U.S. deaths (1). In comparison, unintentional injuries, suicides, and homicides accounted for 7% of mortality, acquired immunodeficiency syndrome for 0.5%, and other infectious diseases for an additional 8%. For many chronic diseases, means of primary, secondary, or tertiary prevention are well known (2,3). It has been estimated that many deaths caused by these six chronic diseases could have been prevented by various means, for example, by effective control of smoking, blood pressure, diet, and alcohol consumption (2,3).
Beginning in January 1989, the MMWR will publish monthly Chronic Disease Reports (CDR) to provide basic information on chronic disease mortality, associated risk factors, and preventive measures. This supplement outlines the contents of information to be published during the first year of CDR; it also notes some limitations, describes the analytic methods used, and discusses applications. Contents of MMWR Chronic Disease Reports
During 1989, nine MMWR CDR will focus on specific chronic diseases; other CDR will present analyses of years of potential life lost, chronic disease mortality trends, and preventable mortality (Table 1). Chronic diseases are defined as diseases that have a prolonged course, that do not resolve spontaneously, and for which a complete cure is rarely achieved, even with treatment. Nine diseases were chosen for the CDR because of their high rates of mortality or their association with known, practical means of prevention. Injuries, occupational diseases, and chronic infectious diseases are not included. The grouping of International Classification of Diseases (ICD) codes in CDR nomenclature reflects shared primary, secondary, or tertiary preventive interventions. CDR nomenclature is not identical to the nomenclature of the World Health Organization (WHO) and the National Center for Health Statistics (NCHS). For example, while NCHS and WHO group male and female breast cancer in one category, "malignant neoplasm of breast," CDR analyze only female breast cancer because secondary preventive measures are known and available for women. For many of these variations, the differences in rates will be small; for example, male breast cancer is a rare disease. In addition, CDR use distinctive terminology to clarify these differences.
Each CDR will provide a table of mortality rates in each state for the featured disease, standardized to the age distribution of the U.S. population in the same year; a map of age-standardized mortality by state will accompany each table. Each report will also include 1) rates of hospitalization for the featured disease in the U.S. population, 2) lists of major modifiable risk factors and preventive measures for that disease, 3) estimates of the prevalence of these risk factors and preventive measures in the U.S. population, and 4) estimates of the crude proportion of each chronic disease in the population attributable to each risk factor and failure to follow each preventive measure. Articles may accompany CDR maps and tables to provide additional information on the etiology, incidence, morbidity, medical and other care, costs, and prevention programs associated with these diseases. Sources of Information in Chronic Disease Reports
NCHS mortality data are based on causes of death reported on death certificates. Autopsy and hospital discharge studies of causes of death noted on certificates have shown a wide range of inaccuracies in death-certificate reporting (4-6). Moreover, relatively few causes of death reported on death certificates are based on autopsy. Reporting of cause of death will not affect estimates of overall mortality, but will affect comparison of mortality due to different causes. In addition, comparison of mortality for different states may be affected by variations among states in diagnostic nomenclature. Where variation is known to occur - for example, in the classification of ischemic heart disease (7) - CDR use broad, inclusive groupings of ICD codes. CDR provide underlying causes of death, defined as "the disease or injury that initiated the train of events leading directly to death or as the circumstances of the accident or violence which produced the fatal injury" (6). Underlying cause of death is determined from causes reported on death certificates, with rules developed by WHO and applied by NCHS (6).
Most death certificates list several causes of death. The likelihood that a specific disease will be selected as the single, underlying cause varies greatly by disease. For example, in 1978 malignant neoplasm of the breast was listed on certificates as cause of death 1.2 times more often than it was reported as the underlying cause of death; diabetes mellitus was listed on certificates 3.9 times as often as reported as underlying cause, and angina pectoris 16.2 times as often (8). The contribution of a particular disease, e.g., diabetes, to mortality may be underestimated by analysis of underlying cause alone. Consequently, the contribution of known risk factors to mortality associated with these diseases may also be underestimated. 2. Hospital discharges
It is useful to assess not only rates of death from chronic diseases, but also the prevalence (i.e., the number of cases of a disease existing in the population at any given time), the incidence (i.e., initial disease occurrence), and the utilization of health-care resources involved in treating patients with these diseases. Information is not always directly available on these characteristics of diseases. However, a rough measure of "disease burden" is provided by information on the discharge diagnoses of hospitalized patients. This information is available in the ongoing National Hospital Discharge Survey (NHDS). CDR present hospital discharge rates for the nine chronic diseases as classified by ICD codes noted in Table 1. Since hospital discharge information is analyzed to estimate incidence, ICD codes selected for CDR analysis are not identical to those selected for analysis of mortality. For example, while an old myocardial infarction (ICD 412) may lead to death, it is not a good indicator of ischemic heart disease incidence; thus, it is omitted from the Hospital Discharge ICD grouping.
NHDS information is based on a national sample of discharges from non-Federal, short-stay hospitals in the United States. Data are abstracted from the summary "face sheet" of medical records. A study of the reliability of this method to assess discharge diagnoses indicated that information from the face sheet was not always adequate to allow valid diagnosis, and that the adequacy of this information varied by disease (9). Thus, reported numbers of hospital discharges should be considered only as approximate indicators of disease occurrence or medical-care use.
While NHDS data provide useful approximations of incidence, they have some limitations. The number of hospital discharges for a given disease does not indicate the number of patients hospitalized, but only the number of hospitalizations for that condition during a set period, usually a year. The number of discharges does not distinguish multiple hospitalizations for one patient from single hospitalizations for multiple patients. Further, NHDS data give no indication of the number of patients with chronic diseases who are not hospitalized because 1) their conditions are not serious enough or are so severe that they die before hospitalization, 2) they have no access to a hospital, or 3) they receive care elsewhere. 3. Population
Information on the population in which exposures, preventive behaviors, morbidity, and mortality occur are based on the 1980 census (10). The Bureau of the Census estimates that the 1980 census overcounted the U.S. population by 0.4% overall (11), thus yielding underestimates of crude mortality rates by the same proportion. Estimated undercounts and overcounts by the census differ by age, race, and sex. For example, in the 1980 census, black males are undercounted by 7.5%, while white and other females are overcounted by 1.7%. Thus, estimated rates of disease events will be artificially high for black males and artificially low for white and other females. The comparison of disease rates by state will be affected insofar as states differ by rates of undercounted persons in various age, race, and sex groups. However, these differences are not considered in CDR.
Estimates of the population for the same year as that for which mortality data are derived are projected from the 1980 census with use of models that incorporate several population characteristics (e.g., births, deaths, migration, military, college, and other institutional associations involving residence away from home) (12). Population counts are available by state and county, sex, age, and race (white and other). 4. Risk factors, preventive measures, and associated relative risks
For each chronic disease, information on risk factors, preventive measures, and the relative risks associated with them is rovided by a panel of experts convened by the Carter Center (2,3). Risk factors chosen for presentation in CDR are those that might be reasonably eliminated or controlled, e.g., hypercholesterolemia, obesity, smoking, and alcohol consumption, and whose eradication or control is not likely to have major adverse effects on health.
Individuals expose themselves to risk factors or practice preventive measures to widely varying degrees. They may not smoke at all, or they may smoke heavily. They may not take any measures to prevent a given disease, or they may take several preventive measures. At a low level of exposure, a given risk factor may have no effect; at a high level, the factor may have no effect beyond that already incurred. Between these levels, risk associated with exposure generally increases with the amount of exposure and decreases with the frequency of preventive activity. However, information on levels of exposure and preventive behavior in the entire U.S. population is not commonly available. For this reason and to simplify analysis, CDR will categorize individuals as being either "exposed" or "unexposed" to a given risk factor and as either "users" or "nonusers" of a preventive behavior at levels that correspond to known risk (or risk reduction) and for which relative-risk estimates are available.
Estimates of the effects of risk factors and preventive measures are derived from studies that vary in design and that may control for different confounding risk factors. The relative risks used for analysis in CDR are chosen to represent the effect of a given exposure on each chronic disease, taking into account other known exposures. Because of different design and control variables, these overall measures of the effect of risk factors and preventive measures are best regarded as approximate.
The effects of risk factors and preventive behaviors on a given disease are not always independent; that is, the effect of one risk factor may be modified (i.e., increased or decreased) in the presence of another risk factor. For example, the effect of asbestos exposure on smokers is greater than the sum of the separate, singular effects of asbestos and of smoking. For such interdependent risk factors, the effects of prevalence of exposure on a gven population will differ to the extent that these risk factors occur simultaneously in individuals. The consequences of multiple interactive risk factors in individuals are not considered in CDR because limited information is available on their population distribution. 5. Prevalence of risk factors and preventive behaviors in the population
Information on the prevalence of risk factors and preventive behaviors in the U.S. population is available from several sources. Other than for smoking-related behaviors (reported by the Office of Smoking and Health, CDC) and alcohol consumption (reported in the Alcohol Epidemiologic Data System by the National Institute on Alcohol Abuse and Alcoholism (NIAAA)), there is no source that provides estimates of the prevalence of risk factors and preventive behaviors in each state. Thus, population-attributable risks (PAR) are calculated using estimates of the national prevalence of various risk factors and preventive behaviors.
The information on the prevalence of risk factors and preventive behaviors is from recent years, principally the last 10 years. However, the exposures that cause or prevent chronic diseases (and death from them) may precede the outcome events by more (or less) than 10 years. To estimate the proportion of deaths attributable to given exposures or failure to use preventive measures, it would be most appropriate to use information on the prevalence of exposures and preventive behaviors in the same populations from earlier periods. However, because the effect of interval since exposure is often unknown and because information on the prevalence of exposure is often unavailable for appropriate periods, CDR present population-attributable risks associated with given exposures, based on the most recent prevalence information available.
Other measures could have been used to compare chronic-disease burdens by state, e.g., number of cases, crude rates, sex- or race-standardized rates, or age-, race-, or sex-specific rates. In CDR standardization by age was chosen because it is a readily comprehensible measure in which a major, inalterable risk factor no longer accounts for differences in disease rates. Thus, differences among state rates may be accounted for, in part, by race and sex differences, as well as by differences in other characteristics such as smoking, diet, alcohol consumption, medical care, and socioeconomic status. b. Population-Attributable Risk
For each chronic disease, it is important to ascertain risk factors that can be eliminated or controlled to reduce the burden of this disease. Similarly, it is important to find measures that can be taken to prevent disease occurrence or to minimize the severity of disease or its consequence, such as death. In a population, the proportion of disease events associated with given risk factors or preventive measures is the PAR. The PAR varies both with the magnitude of the effect of the given risk factor or preventive measure and with the prevalence of the risk factor or preventive behavior in the population. More specifically, PAR = Pe (RR-1) 1 + Pe(RR-1), where Pe is the population prevalence of exposure to the risk factor (or preventive behavior) and RR is the relative risk associated with this risk factor (or preventive behavior). PAR is a proportion and thus must be greater than or equal to 0 and less than or equal to 1. The PAR allows estimation of the number of disease events or deaths that would not have occurred had this risk factor been eliminated in the population, or, in the case of preventive measures, had these measures been appropriately taken in the population at risk (i.e., Number of Prevented Events = Total Number of Events in the Population X PAR). The deaths attributed to different causes may be added only on the assumption that their effects are independent, that is, that the effect of each cause is not affected by the presence or absence of the others. 7. Applications
CDR will provide recent basic information on rates of major preventable chronic diseases in the United States. This information should 1) facilitate priority setting and design of public health programs in chronic disease, 2) provide baseline information for monitoring disease trends and evaluating public health programs, 3) serve as a model for chronic disease surveillance within states (e.g., surveillance of chronic diseases by county), and 4) indicate gaps in existing knowledge. 8. Discussion
Public health attention to chronic diseases in the United States has increased as these diseases have increased in incidence, mortality, and the use of health-care resources. In 1900, tuberculosis, diphtheria, influenza and pneumonia, and various gastrointestinal conditions (most likely infectious) accounted for 38.3% of mortality (14); in 1986, a similar group of conditions caused 3.6% of mortality (1). In 1900, cardiovascular and renal diseases, malignant neoplasms, diabetes, and cirrhosis accounted for 25% of mortality; in 1986, they accounted for 72% of mortality. At CDC, the proportion of MMWR articles devoted to noninfectious diseases has grown from 23% in 1980 to 46% in 1987.
CDR will alert the public health community to recent rates of major preventable chronic diseases in each state and to the principal known means of preventing these diseases and their consequences. Rates and attributed causality presented in CDR will necessarily be approximations. Nevertheless, CDR will serve to inform the public and the public health community about the magnitude and scope of chronic disease in the United States.
Disclaimer All MMWR HTML documents published before January 1993 are electronic conversions from ASCII text into HTML. This conversion may have resulted in character translation or format errors in the HTML version. Users should not rely on this HTML document, but are referred to the original MMWR paper copy for the official text, figures, and tables. An original paper copy of this issue can be obtained from the Superintendent of Documents, U.S. Government Printing Office (GPO), Washington, DC 20402-9371; telephone: (202) 512-1800. Contact GPO for current prices.**Questions or messages regarding errors in formatting should be addressed to firstname.lastname@example.org.
Page converted: 08/05/98
This page last reviewed 5/2/01