Use of a Dose-Response Model to Study Temporal Trends in Spatial Exposure to Coxiella burnetii: Analysis of a Multiyear Outbreak of Q Fever.

The Netherlands underwent a large Q fever outbreak between 2007 and 2009. In this paper, we study spatial and temporal Coxiella burnetii exposure trends during this large outbreak as well as validate outcomes against other published studies and provide evidence to support hypotheses on the causes of the outbreak. To achieve this, we develop a framework using a dose-response model to translate acute Q fever case incidence into exposure estimates. More specifically, we incorporate a geostatistical model that accounts for spatial and temporal correlation of exposure estimates from a human Q fever dose-response model to quantify exposure trends during the outbreak. The 2051 cases, with the corresponding age, gender and residential addresses, reside in the region with the highest attack rates during the outbreak in the Netherlands between 2006 and 2009. We conclude that the multiyear outbreak in the Netherlands is caused by sustained release of infectious bacteria from the same sources, which suggests that earlier implementation of interventions may have prevented many of the cases. The model predicts the risk of infection and acute symptomatic Q fever from multiple exposure sources during a multiple-year outbreak providing a robust, evidence-based methodology to support decision-making and intervention design.

10-year spatial and temporal trends of PM2.5 concentrations in the southeastern US estimated using high-resolution satellite data.

Long-term PM2.5 exposure has been associated with various adverse health outcomes. However, most ground monitors are located in urban areas, leading to a potentially biased representation of true regional PM2.5 levels. To facilitate epidemiological studies, accurate estimates of the spatiotemporally continuous distribution of PM2.5 concentrations are important. Satellite-retrieved aerosol optical depth (AOD) has been increasingly used for PM2.5 concentration estimation due to its comprehensive spatial coverage. Nevertheless, previous studies indicated that an inherent disadvantage of many AOD products is their coarse spatial resolution. For instance, the available spatial resolutions of the Moderate Resolution Imaging Spectroradiometer (MODIS) and the Multiangle Imaging SpectroRadiometer (MISR) AOD products are 10 and 17.6 km, respectively. In this paper, a new AOD product with 1 km spatial resolution retrieved by the multi-angle implementation of atmospheric correction (MAIAC) algorithm based on MODIS measurements was used. A two-stage model was developed to account for both spatial and temporal variability in the PM2.5-AOD relationship by incorporating the MAIAC AOD, meteorological fields, and land use variables as predictors. Our study area is in the southeastern US centered at the Atlanta metro area, and data from 2001 to 2010 were collected from various sources. The model was fitted annually, and we obtained model fitting R2 ranging from 0.71 to 0.85, mean prediction error (MPE) from 1.73 to 2.50 µg m-3, and root mean squared prediction error (RMSPE) from 2.75 to 4.10 µg m-3. In addition, we found cross-validation R2 ranging from 0.62 to 0.78, MPE from 2.00 to 3.01 µgm-3, and RMSPE from 3.12 to 5.00 µgm-3, indicating a good agreement between the estimated and observed values. Spatial trends showed that high PM2.5 levels occurred in urban areas and along major highways, while low concentrations appeared in rural or mountainous areas. Our time-series analysis showed that, for the 10-year study period, the PM2.5 levels in the southeastern US have decreased by ∼20 %. The annual decrease has been relatively steady from 2001 to 2007 and from 2008 to 2010 while a significant drop occurred between 2007 and 2008. An observed increase in PM2.5 levels in year 2005 is attributed to elevated sulfate concentrations in the study area in warm months of 2005.

Estimating vaccine efficacy from outbreak size household data in the presence of heterogeneous transmission probabilities.

We develop a Bayesian approach for estimating vaccine efficacy for susceptibility (VEs) and infectiousness (VEI) using outbreak size household data. Our method allows for heterogeneity in transmission probabilities due to factors that are related to individuals' characteristics, such as age, in addition to vaccination status. It also allows for between-household heterogeneity in transmission probabilities due to random effects associated with households, such as genetic or environmental effects. Using age as a potential covariate causing heterogeneity in individuals' transmission probabilities in households consisting of adults and children, we present the results of a simulation study designed to evaluate the performance of the proposed estimators of VEs and VE(I). We found that estimates of VE(I) have larger bias and variance compared to those of VEs. We also use the approach to compare two vaccination designs: one vaccinating both adults and children, the other only children. Simulations reveal that the design that vaccinates both adults and children provides better estimates of VEs. There is no obvious difference between the two designs in the performance of the estimates of VE(I). In regard to random effects between households and the scenarios considered, models that do not account for between-households heterogeneity produce fairly robust estimates even when household-level random effects are present.

Assessing the role of long-distance translocation and spatial heterogeneity in the raccoon rabies epidemic in Connecticut.

Spatial heterogeneity and long-distance translocation (LDT) play important roles in the spatio-temporal dynamics and management of emerging infectious diseases and invasive species. We assessed the influence of LDT events on the invasive spread of raccoon rabies through Connecticut. We identified several putative LDT events, and developed a network-model to evaluate whether they became new foci for epidemic spread. LDT was fairly common, but many of the LDTs were isolated events that did not spread. Two putative LDT events did appear to become nascent foci that affected the epidemic in surrounding townships. In evaluating the role of LDT, we simultaneously revisited the problem of spatial heterogeneity. The spread of raccoon rabies is associated with forest cover--rabies moves up to three-times slower through the most heavily forested townships compared with those with less forestation. Forestation also modified the effect of rivers. In the best overall model, rabies did not cross the river separating townships that were heavily forested, and the spread slowed substantially between townships that were lightly forested. Our results suggest that spatial heterogeneity can be used to enhance the effects of rabies control by focusing vaccine bait distribution along rivers in lightly forested areas. LDT events are a concern, but this analysis suggests that at a local scale they can be isolated and managed.

Spatiotemporal analysis of epizootic raccoon rabies propagation in Connecticut, 1991-1995.

The quantitative analysis of pathogen transmission within its specific spatial context should improve our ability to predict and control the epizootic spread of that disease. We compared two methods for calibrating the effect of local, spatially distributed environmental heterogeneities on disease spread. Using the time-of-first-appearance of raccoon rabies across the 169 townships in Connecticut, we estimated local spatial variation in township-to-township transmission rate using Trend Surface Analysis (TSA) and then compared these estimates with those based on an earlier probabilistic simulation using the same data. Both the probabilistic simulation and the TSA reveal significant reduction in transmission when local spatial domains are separated by rivers. The probabilistic simulation suggested that township-to-township transmission was reduced sevenfold for townships separated by a river. The global effect of this sevenfold reduction is to increase the time-to-first-appearance in the eastern townships of Connecticut by approximately 29.7% (spread was from west to east). TSA revealed a similar effect of rivers with an overall reduction in rate of local propagation due to rivers of approximately 22%. The 7.7% difference in these two estimates reveals slightly different aspects of the spatial dynamics of this epizootic. Together, these two methods can be used to construct an overall picture of the combined effects of local spatial variation in township-to-township transmission on patterns of local rate of propagation at scales larger than the immediate nearest neighboring townships.

A civil action and statistical assessments of the spatial pattern of disease: do we have a cluster?

A reported "cluster" of excess childhood leukemia cases and possible environmental causes in Woburn, Massachusetts, formed a key motivation for the events described in the popular book and motion picture A Civil Action. Although statistical methods to assess spatial clustering existed prior to the events in Woburn, increasing interest in environmental risk factors and recent developments in geographical information systems and data availability prompt increased attention to such methods and their application to public health data. In this article, we review statistical and epidemiological concepts involved in the analysis of disease clusters. We discuss data issues, outline some methodological approaches, and illustrate ideas using data regarding leukemia incidence in upstate New York for the years 1978-1982.

A comparison of three tests to detect general clustering of a rare disease in Santa Clara County, California.

Statistical tests to detect clustering of a rare disease investigate whether an observed spatial pattern of cases appears to be due to chance alone. Heterogeneous population density and the geographic structure of the data under consideration complicate the ability to make comparisons of different tests. Further, interpretation of test results depends on the nature of the test used and what feature of the data it is designed to detect. With these issues in mind, we compare three recent tests for assessing general clustering among cases where the population is distributed heterogeneously across the study area, namely those of Besag and Newell, Turnbull et al. and Tango. We compare these methods using 1981 incidence data for severe cardiac birth defects from Santa Clara County, California.

Smoking cessation and lung function in mild-to-moderate chronic obstructive pulmonary disease. The Lung Health Study.

Previous studies of lung function in relation to smoking cessation have not adequately quantified the long-term benefit of smoking cessation, nor established the predictive value of characteristics such as airway hyperresponsiveness. In a prospective randomized clinical trial at 10 North American medical centers, we studied 3, 926 smokers with mild-to-moderate airway obstruction (3,818 with analyzable results; mean age at entry, 48.5 yr; 36% women) randomized to one of two smoking cessation groups or to a nonintervention group. We measured lung function annually for 5 yr. Participants who stopped smoking experienced an improvement in FEV(1) in the year after quitting (an average of 47 ml or 2%). The subsequent rate of decline in FEV(1) among sustained quitters was half the rate among continuing smokers, 31 +/- 48 versus 62 +/- 55 ml (mean +/- SD), comparable to that of never-smokers. Predictors of change in lung function included responsiveness to beta-agonist, baseline FEV(1), methacholine reactivity, age, sex, race, and baseline smoking rate. Respiratory symptoms were not predictive of changes in lung function. Smokers with airflow obstruction benefit from quitting despite previous heavy smoking, advanced age, poor baseline lung function, or airway hyperresponsiveness.

Work-related assault injuries among nurses.

Work-related violence is a major public health problem; however, there is a serious deficiency in the knowledge of risk factors for this problem. The purpose of this case-control study was to identify risk factors for work-related assault injuries among nurses. We used unconditional logistic regression to model the dependence of work-related assault injuries on each exposure of interest and the respective confounders. We found a decreased rate for the presence of security personnel (RR = 0.40; 95% CI = 0.19-0.82). We found increased rates for the following factors: the perception that administrators considered assault to be part of the job (RR = 8.14; 95% CI = 3.76-17.60); having received assault prevention training in the current workplace (RR = 4.64; 95% CI = 2.33-9.23); a high (>5) vs. low (<2) patient/personnel ratio (RR = 2.54; 95% CI = 1.13-5.70); working predominantly with patients with mental illness (RR = 3.5; 95% CI = 1.41-8.85); and working with patients who had more than 1- to 4-week and more than 4-week lengths of stay in the institution vs. <1 day (RR = 8.85; 95% CI = 1.58-49.52 and 4.25; 95% CI = 1.17-15.39, respectively).

Environmental justice and statistical summaries of differences in exposure distributions.

Recent regulatory action requires the assessment of environmental justice (equitable protection from the burdens of environmental hazards across sociodemographic subpopulations) in the siting of hazardous waste sites, and prioritization of environmental remediation efforts. Assessments of environmental justice require linking exposure, demographic, and health data. The geographic nature of the data makes the use of geographic information systems attractive for environmental justice assessments. Typical geographic assessments compare the composition of 'exposed' populations, while typical statistical assessments focus on differences in health outcomes between population subgroups, possibly adjusted for exposure. We outline an alternate approach based on summarized differences between exposure distributions within each population subgroup. We illustrate how such summaries provide a tool for site evaluation (e.g., defining exposure inequities resulting from locating a new potential hazard at any of a number of possible sites). In addition, we describe summaries, based on dose-response relationships, to describe risk differences imposed by the observed exposure differences. Reported toxic emissions from Allegheny County, Pennsylvania illustrate the approach.

Discrete distributions for use in twin studies.

We describe several new discrete distributions motivated by the study of longevity in twins. All individuals both living and dead at a given age are randomly paired. The univariate distribution models the number of these pairs where both individuals are alive. If there is a positive association to longevity in twins, then we would expect to see an excessive number of twin pairs both alive at older ages relative to the number of living individuals. We obtain Poisson and normal approximations to the exact distribution. Multivariate distributions are developed to allow for simultaneous and conditional inference at different ages. Odds-ratio parameter models provide a measure of the association of longevity within twin pairs. These models indicate an excessive number of identical twin pairs both alive after age 60 in a cohort of twins born between 1870 and 1880 in Denmark. Monozygotic twins are contrasted with dizygotic twins to separate the genetic and environmental contributions to the similarity in longevity among twins.

Clinical detection of depression among community-based elderly people with self-reported symptoms of depression.

BACKGROUND: Depression is under-diagnosed and under-treated in the primary care sector. The purpose of this study was to determine the association between self-reported indications of depression by community-dwelling elderly enrollees in a managed care organization and clinical detection of depression by primary care clinicians. METHODS: This was a 2-year cohort study of elderly people (n = 3410) who responded to the Geriatric Depression Scale (GDS) at the midpoint of the study period. A broad measure of clinical detection was used consisting of one or more of three indicators: diagnosis of depression, visit to a mental health specialist, or antidepressant medication treatment. RESULTS: Approximately half of the community-based elderly people with self-reported indications of depression (GDS > or = 11) did not have documentation of clinical detection of depression by health providers. Physician recognition of depression tended to increase with the severity of enrollees' self-reported feelings of depression. Men 65-74 years old and those > or = 85 years old were at highest risk for under-detection of depression by primary care providers. CONCLUSIONS: Clinical detection of depression of elderly people living in the community continues to be a problem. The implications of failure to recognize the possibility of depression among elderly White men suggest a serious public health problem.

Log-linear modeling with the negative multinomial distribution.

We develop a negative multinomial sampling plan in which observed cell counts are positively correlated. We show that maximum likelihood estimates of cell means are the same as those found under independent Poisson sampling. There is no maximum likelihood estimate for the shape parameter in general. We propose an estimate of the shape parameter based on the mean and quantiles of Pearson's chi-squared statistic. These techniques are applied to models of cancer incidence for three cities in Ohio and longitudinal health care utilization by a group of senior citizens.

A note on Harold S. Diehl, randomization, and clinical trials.

Harold S. Diehl and coworkers published results from a remarkable trial on the efficacy of vaccines for the common cold in 1938. The original report states that patients were assigned to treatment and control groups "at random." Diehl's study has been referred to as one of the first instances of a randomized, double-blind, placebo-controlled trial. No description of a formal randomization scheme is given in the 1938 report and an unpublished paper of Diehl's suggests the use of alternate assignment in the study.

The analysis of disease clusters, Part II: Introduction to techniques.

Public health professionals often are asked to investigate apparent clusters of human health events or "disease clusters." A cluster is an excess of cases in space (a geographic cluster), in time (a temporal cluster), or in both space and time. This is the second part of an introductory-level review of the analysis of disease clusters for physicians and health professionals concerned with infection surveillance in hospitals. It reviews the status of the field with the hope of expanding the use of cluster analysis methods for the routine surveillance of infectious diseases in the hospital environment.

The analysis of disease clusters, Part I: State of the art.

Public health professionals often are asked to investigate apparent clusters of human health events, or "disease clusters." A cluster is an excess of cases in space (a geographic cluster), in time (a temporal cluster), or in both space and time. This is part I of an introductory-level review of the analysis of disease clusters for physicians and health professionals concerned with infection surveillance in hospitals. It reviews the status of the field with the hope of expanding the use of cluster analysis methods for the routine surveillance of infectious disease in the hospital environment.

Statistical power and design of focused clustering studies.

Focused clustering studies investigate raised incidence of disease in the vicinity of prespecified putative sources of increased risk. The analytic power functions of three focused tests of disease clustering are defined and used to address two design issues related to focused cluster studies. The power functions provide sample sizes required to detect a given increase in relative risk and allow measurement of the effects of aggregating data when a fixed underlying cluster model is assumed. Results are illustrated on hypothetical data as well as leukaemia data from upstate New York.

Disease models implicit in statistical tests of disease clustering.

State and local health departments investigate an increasing number of cluster allegations, for which the selection of appropriate statistical methods is an important problem. Many of the methods for the spatial analysis of health data assume, either implicitly or explicitly, some model of disease occurrence, and comparisons of methods can be difficult when their underlying disease models differ. We review some of the issues involved in the statistical analysis of spatial disease patterns and describe several methods recently proposed to detect areas of increased disease rates. The disease models upon which the methods are based are explicitly described, and they provide a useful basis for comparing alternative clustering methods.

The power of focused tests to detect disease clustering.

Statistical tests have been proposed for determining whether incident cases of adverse health effects are 'clustered' together. Several procedures, termed 'focused', specifically analyse disease surveillance data around pre-specified putative sources of environmental hazard. Little has been done to compare the performance of various proposed methods on actual models of clustering. Analytic power functions are derived for three tests of focused clustering. These functions are based on the probabilistic structure of the clustering tests and do not require simulation. The three tests are compared with respect to statistical power on hypothetical data where monotone multiplicative increases in disease risk near a putative hazard define disease clusters of varying intensity.