Jennifer D. Parker

In 2020 the U.S. Federal Committee on Statistical Methodology (FCSM) released “A Framework for Data Quality”, organized by 11 dimensions of data quality grouped among three domains of quality (utility, objectivity, integrity). This paper addresses the use of the FCSM Framework for data quality assessments of blended data. The FCSM Framework applies to all types of data, however best practices for implementation have not been documented. We applied the FCSM Framework for three health-research related case studies. For each case study, assessments of data quality dimensions were performed to identify threats to quality, possible mitigations of those threats, and trade-offs among them. From these assessments the authors concluded: 1) data quality assessments are more complex in practice than anticipated and expert guidance and documentation are important; 2) each dimension may not be equally important for different data uses; 3) data quality assessments can be subjective and having a quantitative tool could help explain the results, however, quantitative assessments may be closely tied to the intended use of the dataset; 4) there are common trade-offs and mitigations for some threats to quality among dimensions. This paper is one of the first to apply the FCSM Framework to specific use-cases and illustrates a process for similar data uses.

The National Center for Health Statistics is assessing the usefulness of recruited web panels in multiple research areas. One research area examines the use of close-ended probe questions and split-panel experiments for evaluating question-response patterns. Another research area is the development of statistical methodology to leverage the strength of national survey data to evaluate, and possibly improve, health estimates from recruited panels. Recruited web panels, with their lower cost and faster production cycle, in combination with established population health surveys, may be useful for some purposes for statistical agencies. Our initial results indicate that web survey data from a recruited panel can be used for question evaluation studies without affecting other survey content. However, the success of these data to provide estimates that align with those from large national surveys will depend on many factors, including further understanding of design features of the recruited panel (e.g. coverage and mode effects), the statistical methods and covariates used to obtain the original and adjusted weights, and the health outcomes of interest.

This report examines heat-related mortality, cold-related mortality, and other weather-related mortality during 2006-2010 among subgroups of U.S. residents. Weather-related death rates for demographic and area-based subgroups were computed using death certificate information. Adjusted odds ratios for weather-related deaths among subgroups were estimated using logistic regression. During 2006-2010, about 2,000 U.S. residents died each year from weather-related causes of death. About 31% of these deaths were attributed to exposure to excessive natural heat, heat stroke, sun stroke, or all; 63% were attributed to exposure to excessive natural cold, hypothermia, or both; and the remaining 6% were attributed to floods, storms, or lightning. Weather-related death rates varied by age, race and ethnicity, sex, and characteristics of decedent's county of residence (median income, region, and urbanization level). Adjustment for region and urbanization decreased the risk of heat-related mortality among Hispanic persons and increased the risk of cold-related mortality among non-Hispanic black persons, compared with non-Hispanic white persons. Adjustment also increased the risk of heat-related mortality and attenuated the risk of cold-related mortality for counties in the lower three income quartiles. The differentials in weather-related mortality observed among demographic subgroups during 2006-2010 in the United States were consistent with those observed in previous national studies. This study demonstrated that a better understanding of subpopulations at risk from weather-related mortality can be obtained by considering area-based variables (county median household income, region, and urbanization level) when examining weather-related mortality patterns.

Growing evidence has shown the harmful effects of traffic-related pollution on human health, including adverse respiratory, cardiovascular, and pregnancy outcomes. This report describes the linkage of data from the 1999-2008 National Health and Nutrition Examination Surveys (NHANES) and traffic indicators from the 2005 National Highway Planning Network. The residential addresses of NHANES participants were used to assign the distance to the nearest road, the number of roads within concentric buffers of specific radii, and the average annual daily traffic. Summaries of these traffic indicators by participant characteristics, including urbanization of their county of residence, race and ethnicity, poverty status, and health status, were tabulated. Using the traffic indicators, these data show differences in traffic exposure by several participant characteristics including poverty status. Further, reporting of fair or poor health was more common among NHANES respondents nearer to, compared with farther from, roads; this relationship was observed overall and for subgroups defined by urban county of residence, poverty status, and self-reported cigarette smoking. These data may be a resource for understanding relationships between traffic exposure and adverse health, and for identifying subgroups that may be at increased risk. The NHANES-traffic data are restricted use and available to data users in the Research Data Center at the Centers for Disease Control and Prevention's National Center for Health Statistics.

Background: The findings of prior studies of air pollution effects on adverse birth outcomes are difficult to synthesize because of differences in study design. Objectives: The International Collaboration on Air Pollution and Pregnancy Outcomes was formed to understand how differences in research methods contribute to variations in findings. We initiated a feasibility study to a) assess the ability of geographically diverse research groups to analyze their data sets using a common protocol and b) perform location-specific analyses of air pollution effects on birth weight using a standardized statistical approach. Methods: Fourteen research groups from nine countries participated. We developed a protocol to estimate odds ratios (ORs) for the association between particulate matter ≤ 10 μm in aerodynamic diameter (PM10) and low birth weight (LBW) among term births, adjusted first for socioeconomic status (SES) and second for additional location-specific variables. Results: Among locations with data for the PM10 analysis, ORs estimating the relative risk of term LBW associated with a 10-μg/m3 increase in average PM10 concentration during pregnancy, adjusted for SES, ranged from 0.63 [95% confidence interval (CI), 0.30–1.35] for the Netherlands to 1.15 (95% CI, 0.61–2.18) for Vancouver, with six research groups reporting statistically significant adverse associations. We found evidence of statistically significant heterogeneity in estimated effects among locations. Conclusions: Variability in PM10–LBW relationships among study locations remained despite use of a common statistical approach. A more detailed meta-analysis and use of more complex protocols for future analysis may uncover reasons for heterogeneity across locations. However, our findings confirm the potential for a diverse group of researchers to analyze their data in a standardized way to improve understanding of air pollution effects on birth outcomes.

Chemical exposure during pregnancy is potentially harmful to the developing fetus, as the placenta cannot protect against heavy metals such as lead and mercury. Cord blood mercury levels have been associated with childhood cognitive function. High levels of lead exposure during pregnancy have been associated with adverse birth outcomes and, in some studies, with lower cognitive function test scores in childhood; relatively low lead levels have recently been associated with a small risk of decreased birthweight. While intervention for pregnant women with blood lead levels greater than or equal to 5 μg/dL has been recommended, no comparable recommendation for blood mercury in pregnant women has been established.This report presents geometric mean lead and mercury blood levels of pregnant women in the United States based on using the 2003-2008 National Health and Nutrition Examination Surveys (NHANES). Throughout this report, the term mean refers to the geometric mean.

The objective of this study was to evaluate the usefulness of covariates in identifying birth records with implausible values of gestational age. Birthweight distributions for births with early reported gestational ages are markedly bimodal, suggesting a mixture of two distributions. Most births form a normal-shaped left-hand (primary) distribution and a smaller number form the right-hand (secondary) distribution. The births in the secondary distribution are thought to have gestational age mistakenly reported. Prior work has found that births in the secondary distribution are at higher risk of poor outcomes than those in the primary distribution. Using 2002 US Natality data for gestational ages 26-35 weeks, we fit normal mixture models to birthweight with and without covariates (maternal race, education, parity, age, region of the country, prenatal care initiation) by reported gestational age. Additional models were stratified by infant sex. This approach allowed for the relationship between the covariates and birthweight to differ between the components. Mixture models fit reasonably well for reported gestational ages <33 weeks, but not for later weeks. Counter to the hypothesis, results were similar for models with and without covariates or stratification or both, although stratified models without covariates predicted slightly more girls and slightly fewer boys in the secondary distribution than did the corresponding unstratified models. For reported gestational ages <33 weeks, predictions from the four sets of models were highly correlated and predictions were similar for subgroups defined by the clinical estimates of gestational age and other covariates. For births with reported gestational ages of 29 or more weeks, the proportion in the secondary distribution exceeded 30%, although this varied by maternal characteristics. The use of covariates and stratification complicated model fitting without materially improving identification of implausible gestational age values, supporting inferences from prior studies using data 'cleaned' without consideration of maternal or infant characteristics.

For decades, epidemiologists have observed that, among lower birth weight infants, higher risk infants have lower mortality rates than do lower risk infants. However, among higher birth weight infants, the pattern reverses, leading to a riddle of crossing birth weight-specific mortality curves. The riddle has been considered from different perspectives, including relative z scores, directed acyclic graphs, and, most recently, simulated mathematical models of underlying causal factors that produce the observed curves; similarly paradoxical gestational age-specific mortality curves uncross when calculations include all fetuses-at-risk rather than just infants delivered at a particular gestational age. However, researchers have generally focused on birth weight rather than gestational age, likely because birth weight is accurately measured and, if one assumes that birth weight is an intermediate variable between the underlying causal factors and mortality, is easier to model. Within the framework of existing analytical approaches, adding the complexity of a direct relation between gestational age and mortality, and possibly more complex relations among the casual factors, may be difficult. Nevertheless, duration of pregnancy seems a better proxy for the true construct of interest, whether the baby is mature enough to survive, so shifting attention to understanding the riddle of gestational age-specific mortality is encouraged.

BackgroundChildhood respiratory allergies, which contribute to missed school days and other activity limitations, have increased in recent years, possibly due to environmental factors.ObjectiveIn this study we examined whether air pollutants are associated with childhood respiratory allergies in the United States.MethodsFor the approximately 70,000 children from the 1999–2005 National Health Interview Survey eligible for this study, we assigned between 40,000 and 60,000 ambient pollution monitoring data from the U.S. Environmental Protection Agency, depending on the pollutant. We used monitors within 20 miles of the child’s residential block group. We used logistic regression models, fit with methods for complex surveys, to examine the associations between the reporting of respiratory allergy or hay fever and annual average exposure to particulate matter ≤ 2.5 μm in diameter (PM2.5), PM ≤ 10 μm in diameter, sulfur dioxide, and nitrogen dioxide and summer exposure to ozone, controlling for demographic and ...

This report describes the linked data file obtained as a result of combining air pollution data and National Health and Nutrition Examination Survey (NHANES) III data. Average annual air pollution exposures to particulate matter consisting of particles smaller than 10 micrometers in diameter (PM10), sulfur dioxide (SO2), nitrogen dioxide (NO2), and carbon monoxide (CO) were created for NHANES III examined persons by averaging values from monitors within a 5-, 10-, 15-, and 20-mile radius from the block-group centroid of their residence and in the county of their residence. Percentage records geocoded to block-group level, percentage records linked to air pollution, and distributions of exposure values were estimated for the total sample and various demographic groups. The percentages of respondents who were assigned countywide air pollution values ranges from a low of 43 percent in the case of NO2 data to a high of 68 percent in the case of PM10 data. Among the pollutants considered, PM10 data provides the best coverage. Of all the metrics created, the highest coverage is achieved by averaging readings of monitors located within a 20-mile distance from the centroid of respondents' block groups. Among the demographic variables analyzed, differences in air pollution coverage and exposure levels occur most often among groups defined by race and Hispanic origin, region, and county level of urbanization. However, differences among groups depend on the pollutant and geographic linkage method. The linked dataset provides researchers with opportunities to investigate the relationship between air pollution and various health outcomes.