Search

This paper examines effect modification of heat- and cold-related mortality in seven US cities in 1986-1993. City-specific Poisson regression analyses of daily noninjury mortality were fit with predictors of mean daily apparent temperature (a construct reflecting physiologic effects of temperature and humidity), time, barometric pressure, day of the week, and particulate matter less than 10 micro m in aerodynamic diameter. Percentage change in mortality was calculated at 29 degrees C apparent temperature (lag 0) and at -5 degrees C (mean of lags 1, 2, and 3) relative to 15 degrees C. Separate models were fit to death counts stratified by age, race, gender, education, and place of death. Effect estimates were combined across cities, treating city as a random effect. Deaths among Blacks compared with Whites, deaths among the less educated, and deaths outside a hospital were more strongly associated with hot and cold temperatures, but gender made no difference. Stronger cold associations were found for those less than age 65 years, but heat effects did not vary by age. The strongest effect modifier was place of death for heat, with out-of-hospital effects more than five times greater than in-hospital deaths, supporting the biologic plausibility of the associations. Place of death, race, and educational attainment indicate vulnerability to temperature-related mortality, reflecting inequities in health impacts related to climate change.

While time-series studies have consistently provided evidence for an effect of particulate air pollution on mortality, uncertainty remains as to the extent of the life-shortening implied by those associations. In this paper, the authors estimate the association between air pollution and mortality using different timescales of variation in the air pollution time series to gain further insight into this question. The authors' method is based on a Fourier decomposition of air pollution time series into a set of independent exposure variables, each representing a different timescale. The authors then use this set of variables as predictors in a Poisson regression model to estimate a separate relative rate of mortality for each exposure timescale. The method is applied to a database containing information on daily mortality, particulate air pollution, and weather in four US cities (Pittsburgh, Pennsylvania; Minneapolis, Minnesota; Seattle, Washington; and Chicago, Illinois) from the period 1987-1994. The authors found larger relative rates of mortality associated with particulate air pollution at longer timescale variations (14 days-2 months) than at shorter timescales (1-4 days). These analyses provide additional evidence that associations between particle indexes and mortality do not imply only an advance in the timing of death by a few days for frail individuals.

The pattern of deterioration in patients with Alzheimer's disease is highly variable within a given population. With recent speculation that the apolipoprotein E allele may influence rate of decline and claims that certain drugs may slow the course of the disease, there is a compelling need for sound statistical methodology to address these questions. Current statistical methods for describing decline do not adequately take into account between-patient variability and possible floor and/or ceiling effects in the scale measuring decline, and they fail to allow for uncertainty in disease onset. In this paper, the authors analyze longitudinal Mini-Mental State Examination scores from two groups of Alzheimer's disease subjects from Palo Alto, California, and Minneapolis, Minnesota, in 1981-1993 and 1986-1988, respectively. A Bayesian hierarchical model is introduced as an elegant means of simultaneously overcoming all of the difficulties referred to above.

A population-based case-control study of renal cell cancer was conducted in Minnesota between 1988 and 1990. It included 690 histologically confirmed incident cases identified through the state cancer surveillance system, and 707 age and sex frequency-matched controls. In this paper, the authors present estimates of the proportion of renal cell cancer cases attributable (or population attributable risk (PAR)) to 1) well-established risk factors, namely smoking, excess weight, and hypertension, and 2) more speculative risk factors, namely elevated protein intake, history of renal disease (i.e., stone, injury, or infection) and high parity among women. These estimates were based on information obtained from directly interviewed subjects (65% of cases and 100% of controls). The PARs for the three main risk factors were 21% for hypertension (defined by a reported personal history of hypertension or of treatment with antihypertensive or diuretic drugs), 21% for excess weight (defined by an elevation of the usual body mass index above the first quartile), and 18% for smoking (past and current). Overall, these three factors accounted for 49% of cases. The proportion increased to 60% when the three more speculative risk factors were considered as well. Sex-specific analyses revealed a greater impact of smoking among men mainly due to a higher prevalence of past smoking. In contrast, the impact of hypertension and excess weight was greater among women due to higher relative risks. These results suggest that 1) intervention measures aimed at reducing smoking, excess weight, and hypertension could substantially lower the overall incidence of renal cell cancer, and 2) intervention measures may have to be sex-specific. Conversely, because at least 40% of cases remain unexplained by the risk factors under study, further etiologic research is needed into renal cell cancer, an increasingly common form of cancer.

The authors present a method of estimating the duration of ongoing prevention trials, showing how the method was applied to the Colon Cancer Control Study, a University of Minnesota study of occult blood testing undertaken to screen for colorectal cancer in older Minnesotans. In that study, begun in 1975 and ongoing, as in several other recent large prevention trials, it was necessary, after the start of the study, to revise upward the initial estimates of study duration derived from general population experience. The underestimates arose because the planners had not adequately taken various population selection factors into account in estimating the expected number of events in the control group. In this paper, the authors outline a method for estimating control group event rates and study duration requirements (and, in some circumstances, also sample size requirements) of prevention studies, via models of disease-specific and all-cause standardized mortality ratios which adjust for various selection effects. The authors also validate the model for disease-specific standardized mortality ratios by means of independent estimates of disease incidence and case survival.

Prior to 1979, epidemiologic studies which included mortality follow-up on large cohorts relied on death certificates from last state of residence and expensive follow-up techniques to determine survival. Beginning with 1979, the National Death Index can be used to search death certificate files from all 50 states, the Virgin Islands, and Puerto Rico. This paper addresses the issue of whether mortality follow-up in epidemiologic studies based on a single state death certificate search using only data available in 1970-1975 can be compared with post-1979 mortality follow-up using the National Death Index. This question was addressed by following a cohort of 2,925 coronary heart disease and cerebrovascular disease 1980 hospital discharges from 1980 through 1983 with the use of both the National Death Index and the Minnesota Death Index (MINNDEX). Algorithms for evaluating potential death certificate matches were developed independently for both systems. The systems agreed on the survival status of 98% of the cohort, and both identified 31% as dead. This study supports the comparison of results of National Death Index follow-up to pre-National Death Index studies using algorithms such as the MINNDEX, and provides evidence that trend analyses relying on single state death searches pre-1979 and on the National Death Index from 1979 are valid, particularly in chronically ill persons.