Methodology of studies evaluating death certificate accuracy were flawed.

BACKGROUND AND OBJECTIVE: Statistics on causes of death are important for epidemiologic research. Studies that evaluate the source data often give conflicting results, which raise questions about comparability and validity of methods. METHODS: For 44 recent evaluation studies we examined the methods employed and assessed the reproducibility. RESULTS: Thirty studies stated who reviewed the source data. Six studies reported reliability tests. Twelve studies included all causes of death, but none specified criteria for identifying the underlying cause when several, etiologically independent conditions were present. We assessed these as not reproducible. Of 32 studies that focussed on a specific condition, 21 provided diagnostic criteria such that the verification of the focal diagnosis is reproducible. Of 16 that discussed the difference between dying "with" and "from" a condition, eight described how competing causes had been handled. For these eight, the selection of a principal cause is reproducible, but in three the selection strategy conflicts with the international instructions issued by the World Health Organization. CONCLUSION: Methods and criteria are often insufficiently described. When described, they sometimes disagree with the international standard. Explicit descriptions of methods and criteria would contribute to methodologic improvement and would allow readers to assess the generalizability of the conclusions.

Estimating deaths attributable to obesity in the United States.

Estimates of deaths attributable to obesity in the United States rely on estimates from epidemiological cohorts of the relative risk of mortality associated with obesity. However, these relative risk estimates are not necessarily appropriate for the total US population, in part because of exclusions to control for baseline health status and exclusion or underrepresentation of older adults. Most deaths occur among older adults; estimates of deaths attributable to obesity can vary widely depending on the assumptions about the relative risks of mortality associated with obesity among the elderly. Thus, it may be difficult to estimate deaths attributable to obesity with adequate accuracy and precision. We urge efforts to improve the data and methods for estimating this statistic.

Disease classification: measuring the effect of the Tenth Revision of the International Classification of Diseases on cause-of-death data in the United States.

The purpose of this paper is to describe the statistical impact of the Tenth Revision of the International Classification of Diseases (ICD-10) on cause-of-death data for the United States. ICD-10 was implemented in the U.S. effective with deaths occurring in 1999. The paper is based on cause-of-death information from a large sample of 1996 death certificates filed in the 50 States and the District of Columbia. Cause-of-death information in the sample includes underlying cause of death classified by both ICD-9 and ICD-10. Preliminary comparability ratios by cause of death presented in this paper indicate the extent of discontinuities in cause-of-death trends from 1998 to 1999 resulting from implementing ICD-10. For some leading causes (for example, septicaemia, influenza and pneumonia, Alzheimer's disease, and nephritis, nephrotic syndrome and nephrosis) the discontinuity in trend is substantial. Results of this study, although preliminary, are essential to analysing trends in mortality statistics between ICD-9 and ICD-10. In particular, the results provide a means for interpreting changes between 1998, which is the last year in which ICD-9 was used, and 1999, the year in which ICD-10 was implemented for mortality in the United States. Published in 2003 by John Wiley & Sons, Ltd.

Increase in deaths caused by HIV infection due to changes in rules for selecting underlying cause of death.

With implementation of the (ICD-10), for U.S. vital statistics in 1999, the criteria for selecting HIV infection as the underlying cause of death were expanded. To estimate the effect of ICD-10 rules on the number of deaths attributed to HIV infection, we applied a simplified version of ICD-10 rules to data on causes of death from all U.S. death certificates for 1998 (previously classified by rules of the 9th revision of ICD [ICD-9]) and calculated the resulting increase in deaths for which HIV infection was selected as the underlying cause. Of the 17,186 deaths in 1998 with any mention of HIV infection on the death certificate, ICD-10 rules selected HIV infection as the underlying cause for 15,145, which was 1,719 (13%) more than the 13,426 for which it had been selected by ICD-9 rules. The proportional increase differed by demographic group, being less among non-Hispanic blacks (9%) and Hispanics (13%) than among non-Hispanic whites (18%). Thus, comparison of deaths attributed to HIV infection in 1999 or later with those in 1998 or earlier should take into account the changes in ICD rules for selecting the underlying cause of death.

Annual report to the nation on the status of cancer, 1973-1999, featuring implications of age and aging on U.S. cancer burden.

BACKGROUND: The American Cancer Society, the National Cancer Institute, the North American Association of Central Cancer Registries (NAACCR), the National Institute on Aging (NIA), and the Centers for Disease Control and Prevention, including the National Center for Health Statistics (NCHS) and the National Center for Chronic Disease Prevention and Health Promotion, collaborated to provide an annual update on cancer occurrence and trends in the United States. This year's report contained a special feature focusing on implications of age and aging on the U.S. cancer burden. METHODS: For 1995 through 1999, age-specific rates and age-adjusted rates were calculated for the major cancers using incidence data from the Surveillance, Epidemiology, and End Results Program, the National Program of Cancer Registries, and the NAACCR, and mortality data from NCHS. Joinpoint analysis, a model of joined line segments, was used to examine 1973-1999 trends in incidence and death rates by age for the four most common cancers. Deaths were classified using the eighth, ninth, and tenth revisions of the International Classification of Diseases. Age-adjusted incidence and death rates were standardized to the year 2000 population, which places more emphasis on older persons, in whom cancer rates are higher. RESULTS: Across all ages, overall cancer death rates decreased in men and women from 1993 through 1999, while cancer incidence rates stabilized from 1995 through 1999. Age-specific trends varied by site, sex, and race. For example, breast cancer incidence rates increased in women aged 50-64 years, whereas breast cancer death rates decreased in each age group. However, a major determinant of the future cancer burden is the demographic phenomenon of the aging and increasing size of the U.S. population. The total number of cancer cases can be expected to double by 2050 if current incidence rates remain stable. CONCLUSIONS: Despite the continuing decrease in cancer death rates and stabilization of cancer incidence rates, the overall growth and aging of the U.S. population can be expected to increase the burden of cancer in our nation.