Commentary: On measurement error, PSA doubling time, and prostate cancer.

Exposure measurement error is a pervasive problem for epidemiology research projects designed to provide valid and precise statistical evidence supporting postulated exposure-disease relationships of interest. The purpose of this commentary is to highlight an important real-life example of this exposure measurement error problem and to provide a simple and useful diagnostic tool for physicians and their patients that corrects for the exposure measurement error. More specifically, prostate-specific antigen doubling time (PSADT) is a widely used measure for guiding future treatment options for patients with biochemically recurrent prostate cancer. Numerous papers have been published claiming that a low calculated PSADT value (denoted PSADT^) is predictive of metastasis and premature death from prostate cancer. Unfortunately, none of these papers have adjusted for the measurement error in PSADT^, an estimator that is typically computed using the popular Memorial Sloan Kettering website very often visited by both physicians and their patients. For this website, the estimator PSADT^ of the true (but unknown) PSADT for a patient (denoted PSADT∗) is computed as the natural log of 2 (i.e., 0.6931) divided by the estimated slope of the straight-line regression of the natural log of PSA (in ng/mL) on time. We utilize PSADT^ to derive an expression for the probability that the unknown PSADT∗ for a patient is below a specified value C (>0) of concern to both the physician and the patient. This probability is easy to interpret and takes into account the fact that PSADT^ is a statistical estimator with variability. This variability introduces measurement error, namely, the difference between a computed value PSADT^ and the true, but unknown, value PSADT∗. We have developed an Excel calculator that, once the [time, ln(PSA)] values are entered, outputs both the value of PSADT^ and the desired probability. In addition, we discuss problematic statistical issues attendant with PSADT∗ estimation typically based on at most three or four PSA values. We strongly recommend the use of this probability when physicians are discussing PSADT^ values and associated treatment options with their patients. And, we stress that future epidemiology research projects involving PSA doubling time should take into account the measurement error problem highlighted in this Commentary.

Starting the Conversation: Are Campus Sexual Assault Policies Related to the Prevalence of Campus Sexual Assault?

One goal of university campus sexual assault (CSA) policies is to help prevent CSA. Federal guidance in the 2014 White House Task Force to Protect Students From Sexual Assault Checklist for Campus Sexual Misconduct Policies suggests 10 elements for inclusion in CSA policies (e.g., Policy Introduction, Grievance/Adjudication), and outlines policy topics to be included within each element (Policy Introduction includes two topics: statement of prohibition against sex discrimination including sexual misconduct and statement of commitment to address sexual misconduct). However, no research has examined whether CSA policies impact CSA prevalence. To begin addressing this gap, we studied 24 universities participating in the 2015 Association of American Universities Campus Climate Survey on Sexual Assault and Sexual Misconduct. We linked 2014-2015 data from these universities' CSA policies and their CSA prevalence findings from the 2015 Association of American Universities (AAU) survey. To test whether the comprehensiveness of schools' CSA policies was related to schools' CSA prevalence, we examined the degree to which the CSA policies included recommended policy content from the aforementioned Checklist. Policies were characterized as more comprehensive if they included greater numbers of Checklist topics. We then correlated the number of topics within the policies with school-level CSA prevalence. We also explored whether there was lower CSA prevalence among schools with policies containing particular topics. Results suggested that greater comprehensiveness of schools' entire CSA policies was negatively correlated with CSA prevalence; however, these findings did not approach statistical significance. The number of negative correlations observed between schools' CSA policy elements and CSA prevalence among undergraduate women was greater than expected by chance alone, suggesting a possible connection between comprehensive CSA policies and CSA prevalence. Schools with policies that included a topic on their sexual assault response team had the lowest CSA prevalence for both women and men, and schools that included topics describing grievance/adjudication procedures had lower CSA prevalence. This study provides a novel examination of CSA and could inform needed research related to the impact of CSA policies on CSA.

Numeric score-based conditional and overall change-in-status indices for ordered categorical data.

Planned interventions and/or natural conditions often effect change on an ordinal categorical outcome (e.g., symptom severity). In such scenarios, it is sometimes desirable to assign a priori scores to observed changes in status, typically giving higher weight to changes of greater magnitude. We define change indices for such data based upon a multinomial model for each row of a c × c table, where the rows represent the baseline status categories. We distinguish an index designed to assess conditional changes within each baseline category from two others designed to capture overall change. One of these overall indices measures expected change across a target population. The other is scaled to capture the proportion of total possible change in the direction indicated by the data, so that it ranges from -1 (when all subjects finish in the least favorable category) to +1 (when all finish in the most favorable category). The conditional assessment of change can be informative regardless of how subjects are sampled into the baseline categories. In contrast, the overall indices become relevant when subjects are randomly sampled at baseline from the target population of interest, or when the investigator is able to make certain assumptions about the baseline status distribution in that population. We use a Dirichlet-multinomial model to obtain Bayesian credible intervals for the conditional change index that exhibit favorable small-sample frequentist properties. Simulation studies illustrate the methods, and we apply them to examples involving changes in ordinal responses for studies of sleep deprivation and activities of daily living.

Assortativity coefficient-based estimation of population patterns of sexual mixing when cluster size is informative.

OBJECTIVES: Population sexual mixing patterns can be quantified using Newman's assortativity coefficient (r). Suggested methods for estimating the SE for r may lead to inappropriate statistical conclusions in situations where intracluster correlation is ignored and/or when cluster size is predictive of the response. We describe a computer-intensive, but highly accessible, within-cluster resampling approach for providing a valid large-sample estimated SE for r and an associated 95% CI. METHODS: We introduce needed statistical notation and describe the within-cluster resampling approach. Sexual network data and a simulation study were employed to compare within-cluster resampling with standard methods when cluster size is informative. RESULTS: For the analysis of network data when cluster size is informative, the simulation study demonstrates that within-cluster resampling produces valid statistical inferences about Newman's assortativity coefficient, a popular statistic used to quantify the strength of mixing patterns. In contrast, commonly used methods are biased with attendant extremely poor CI coverage. Within-cluster resampling is recommended when cluster size is informative and/or when there is within-cluster response correlation. CONCLUSIONS: Within-cluster resampling is recommended for providing valid statistical inferences when applying Newman's assortativity coefficient r to network data.

Approximate and Pseudo-Likelihood Analysis for Logistic Regression Using External Validation Data to Model Log Exposure.

A common goal in environmental epidemiologic studies is to undertake logistic regression modeling to associate a continuous measure of exposure with binary disease status, adjusting for covariates. A frequent complication is that exposure may only be measurable indirectly, through a collection of subject-specific variables assumed associated with it. Motivated by a specific study to investigate the association between lung function and exposure to metal working fluids, we focus on a multiplicative-lognormal structural measurement error scenario and approaches to address it when external validation data are available. Conceptually, we emphasize the case in which true untransformed exposure is of interest in modeling disease status, but measurement error is additive on the log scale and thus multiplicative on the raw scale. Methodologically, we favor a pseudo-likelihood (PL) approach that exhibits fewer computational problems than direct full maximum likelihood (ML) yet maintains consistency under the assumed models without necessitating small exposure effects and/or small measurement error assumptions. Such assumptions are required by computationally convenient alternative methods like regression calibration (RC) and ML based on probit approximations. We summarize simulations demonstrating considerable potential for bias in the latter two approaches, while supporting the use of PL across a variety of scenarios. We also provide accessible strategies for obtaining adjusted standard errors to accompany RC and PL estimates.

Somatic symptoms among US adolescent females: associations with sexual and physical violence exposure.

The objective of this study is to examine the association between physical and sexual violence exposure and somatic symptoms among female adolescents. We studied a nationally representative sample of 8,531 females, aged 11-21 years, who participated in the 1994-1995 Wave I of the National Longitudinal Study of Adolescent Health (Add Health). Female adolescents were asked how often they had experienced 16 specific somatic symptoms during the past 12 months. Two summary categorical measures were constructed based on tertiles of the distributions for the entire female sample: (a) total number of different types of symptoms experienced, and (b) number of frequent (once a week or more often) different symptoms experienced. Groups were mutually exclusive. We examined associations between adolescents' violence exposure and somatic symptoms using multinomial logistic regression analyses. About 5 % of adolescent females reported both sexual and non-sexual violence, 3 % reported sexual violence only, 36 % reported non-sexual violence only, and 57 % reported no violence. Adolescents who experienced both sexual and non-sexual violence were the most likely to report many different symptoms and to experience very frequent or chronic symptoms. Likelihood of high symptomatology was next highest among adolescents who experienced sexual violence only, followed by females who experienced non-sexual violence only. Findings support an exposure-response association between violence exposure and somatic symptoms, suggesting that symptoms can be markers of victimization. Treating symptoms alone, without addressing the potential violence experienced, may not adequately improve adolescents' somatic complaints and well-being.

Intimate partner violence and depressive symptoms before pregnancy, during pregnancy, and after infant delivery: an exploratory study.

Women who experience intimate partner violence (IPV) during pregnancy also tend to experience depressive symptoms. Unfortunately, little is known about how victimized women's levels of depressive symptoms change longitudinally before pregnancy, during pregnancy, and after infant delivery. In addition, few studies have used a comparison group of women to determine if levels of depressive symptoms among victimized women differ from depressive symptom levels in women who have not experienced IPV. To help address these knowledge gaps, we examined longitudinal trends in levels of depressive symptoms among a sample of 76 women who did (n = 33) and did not (n = 43) experience physical IPV during pregnancy. Using multilevel analysis, we estimated the relationship of physical IPV victimization and women's depressive symptom levels across six time periods: (a) the year before pregnancy, (b) first and second trimesters, (c) third trimester, (d) the first month postpartum, (e) Months 2 to 6 postpartum, and (f) Months 7 to 12 postpartum. Women who experienced physical IPV victimization during pregnancy had significantly higher levels of depressive symptoms during each time period (p < .05). No significant difference between the two groups was found in the rate of change in levels of depressive symptoms over time. These findings point to the importance of screening for IPV within health care settings and suggest that women physically abused during pregnancy need safety interventions that are coordinated with interventions targeting symptoms of depression.

Low-dose metabolism of benzene in humans: science and obfuscation.

Benzene is a ubiquitous air pollutant that causes human leukemia and hematotoxic effects. Although the mechanism by which benzene causes toxicity is unclear, metabolism is required. A series of articles by Kim et al. used air and biomonitoring data from workers in Tianjin, China, to investigate the dose-specific metabolism (DSM) of benzene over a wide range of air concentrations (0.03-88.9 p.p.m.). Kim et al. concluded that DSM of benzene is greatest at air concentrations <1 p.p.m. This provocative finding motivated the American Petroleum Institute to fund a study by Price et al. to reanalyze the original data. Although their formal 'reanalysis' reproduced Kim's finding of enhanced DSM at sub-p.p.m. benzene concentrations, Price et al. argued that Kim's methods were inappropriate for assigning benzene exposures to low exposed subjects (based on measurements of urinary benzene) and for adjusting background levels of metabolites (based on median values from the 60 lowest exposed subjects). Price et al. then performed uncertainty analyses under alternative approaches, which led them to conclude that '… the Tianjin data appear to be too uncertain to support any conclusions …' regarding the DSM of benzene. They also argued that the apparent low-dose metabolism of benzene could be explained by 'lung clearance.' In addressing these criticisms, we show that the methods and arguments presented by Price et al. are scientifically unsound and that their results are unreliable.

GRADE guidelines: 13. Preparing summary of findings tables and evidence profiles-continuous outcomes.

Presenting continuous outcomes in Summary of Findings tables presents particular challenges to interpretation. When each study uses the same outcome measure, and the units of that measure are intuitively interpretable (e.g., duration of hospitalization, duration of symptoms), presenting differences in means is usually desirable. When the natural units of the outcome measure are not easily interpretable, choosing a threshold to create a binary outcome and presenting relative and absolute effects become a more attractive alternative. When studies use different measures of the same construct, calculating summary measures requires converting to the same units of measurement for each study. The longest standing and most widely used approach is to divide the difference in means in each study by its standard deviation and present pooled results in standard deviation units (standardized mean difference). Disadvantages of this approach include vulnerability to varying degrees of heterogeneity in the underlying populations and difficulties in interpretation. Alternatives include presenting results in the units of the most popular or interpretable measure, converting to dichotomous measures and presenting relative and absolute effects, presenting the ratio of the means of intervention and control groups, and presenting the results in minimally important difference units. We outline the merits and limitations of each alternative and provide guidance for meta-analysts and guideline developers.

Are pregnant and postpartum women: at increased risk for violent death? Suicide and homicide findings from North Carolina.

The purpose of this study is to estimate rates of suicide and homicide death among pregnant, postpartum and non-pregnant/non-postpartum women ages 14-44, and to determine comparative rates of violent death for pregnant and/or postpartum women compared to non-pregnant/non-postpartum women. North Carolina surveillance and vital statistics data from 2004 to 2006 were used to examine whether pregnant or postpartum women have higher (or lower) rates of suicide and homicide compared to other reproductive-aged women. The suicide rate for pregnant women was 27% of the rate for non-pregnant/non-postpartum women (rate ratio= 0.27, 95% CI = 0.11-0.66), and the suicide rate for postpartum women was 54% of the rate for non-pregnant/non-postpartum women (rate ratio = 0.54, 95% CI = 0.31-0.95). Homicide rates also were lower for pregnant and postpartum women, with the homicide rate for pregnant women being 73% of the rate for non-pregnant/non-postpartum women (rate ratio = 0.73, 95% CI = 0.39-1.37), and the homicide rate for postpartum women being half the rate for non-pregnant/non-postpartum women (rate ratio = 0.50, 95% CI = 0.26-0.98). Although pregnant and postpartum women are at risk for homicide and suicide death, the highest risk group is non-pregnant/non-postpartum women. Violence prevention efforts should target all women of reproductive age, and pay particular attention to non-pregnant/non-postpartum women, who may have less access to health care services than pregnant and postpartum women.

Suicide and homicide in North Carolina: Initial findings from the North Carolina Violent Death Reporting System, 2004-2007.

BACKGROUND: Research concerning suicide and homicide in North Carolina is needed so that medical providers and others who develop and implement preventive and therapeutic interventions related to violence have an empirical base from which to work. METHODS: North Carolina Violent Death Reporting System data composed of death certificates, medical examiner reports, and law enforcement reports were analyzed to examine the prevalence of suicide and homicide in North Carolina during 2004-2007 and to describe the sociodemographic characteristics of suicide and homicide victims. RESULTS: Suicides and homicides accounted for 2.3% of all North Carolina deaths during 2004-2007. There were 12.0 suicides (95% confidence interval [CI], 11.7-12.4) and 7.2 homicides (95% CI, 6.9-7.4) per 100,000 North Carolina residents. Suicide rates were higher among men and boys, whites, non-Hispanics, and persons aged > or = 35 years. Homicide rates were higher among men and boys, American Indians, blacks, Hispanics, and persons aged < or = 24 years. Firearms were the most common method used to commit suicide and homicide, accounting for 59.5% of suicides and 67.0% of homicides. CONCLUSIONS: Every day in North Carolina, approximately 3 persons kill themselves and approximately 2 persons are killed by others. Suicide and homicide inflict a high level of preventable mortality in North Carolina. Learning more about these violent deaths will help to inform the development of effective violence-prevention interventions.

Patterns of intimate partner violence victimization from adolescence to young adulthood in a nationally representative sample.

PURPOSE: To determine the prevalence of patterns of intimate partner violence (IPV) victimization from adolescence to young adulthood, and document associations with selected sociodemographic and experiential factors. METHODS: We used prospective data from the National Longitudinal Study of Adolescent Health to group 4134 respondents reporting only opposite-sex romantic or sexual relationships in adolescence and young adulthood into four victimization patterns: no IPV victimization, adolescent-limited IPV victimization, young adult onset IPV victimization, and adolescent-young adult persistent IPV victimization. RESULTS: Forty percent of respondents reported physical or sexual victimization by young adulthood. Eight percent experienced IPV only in adolescence, 25% only in young adulthood, and 7% showed persistent victimization. Female sex, Hispanic and non-Hispanic black race/ethnicity, an atypical family structure (something other than two biologic parents, step-family, single parent), more romantic partners, experiencing childhood abuse, and early sexual debut (before age 16) were each associated with one or more patterns of victimization versus none. Number of romantic partners and early sexual debut were the most consistent predictors of violence, its timing of onset, and whether victimization persisted across developmental periods. These associations did not vary by biological sex. CONCLUSIONS: Substantial numbers of young adults have experienced physical or sexual IPV victimization. More research is needed to understand the developmental and experiential mechanisms underlying timing of onset of victimization, whether victimization persists across time and relationships, and whether etiology and temporal patterns vary by type of violence. These additional distinctions would inform the timing, content, and targeting of violence prevention efforts.

Investigation of PAH biomarkers in the urine of workers exposed to hot asphalt.

Airborne emissions from hot asphalt contain mixtures of polycyclic aromatic hydrocarbons (PAHs), including several carcinogens. We investigated urinary biomarkers of three PAHs, namely naphthalene (Nap), phenanthrene (Phe), and pyrene (Pyr) in 20 road-paving workers exposed to hot asphalt and in 6 road milling workers who were not using hot asphalt (reference group). Our analysis included baseline urine samples as well as postshift, bedtime, and morning samples collected over three consecutive days. We measured unmetabolized Nap (U-Nap) and Phe (U-Phe) as well as the monohydroxylated metabolites of Nap (OH-Nap), Phe (OH-Phe), and Pyr (OH-Pyr) in each urine sample. In baseline samples, no significant differences in biomarker levels were observed between pavers and millers, suggesting similar background exposures. In postshift, bedtime, and morning urine samples, the high pairwise correlations observed between levels of all biomarkers suggest common exposure sources. Among pavers, levels of all biomarkers were significantly elevated in postshift samples, indicating rapid uptake and elimination of PAHs following exposure to hot asphalt (biomarker levels were not elevated among millers). Results from linear mixed-effects models of levels of U-Nap, U-Phe, OH-Phe, and OH-Pyr across pavers showed significant effects of work assignments with roller operators having lower biomarker levels than the other workers. However, no work-related effect was observed for levels of OH-Nap, apparently due to the influence of cigarette smoking. Biological half-lives, estimated from regression coefficients for time among pavers, were 8 h for U-Phe, 10 h for U-Nap, 13 h for OH-Phe and OH-Pyr, and 26 h for OH-Nap. These results support the use of U-Nap, U-Phe, OH-Phe, and OH-Pyr, but probably not OH-Nap, as short-term biomarkers of exposure to PAHs emanating from hot asphalt.

Comparing urinary biomarkers of airborne and dermal exposure to polycyclic aromatic compounds in asphalt-exposed workers.

When working with hot mix asphalt, road pavers are exposed to polycyclic aromatic hydrocarbons (PAHs) through the inhalation of vapors and particulate matter (PM) and through dermal contact with PM and contaminated surfaces. Several PAHs with four to six rings are potent carcinogens which reside in these particulate emissions. Since urinary biomarkers of large PAHs are rarely detectable in asphalt workers, attention has focused upon urinary levels of the more volatile and abundant two-ring and three-ring PAHs as potential biomarkers of PAH exposure. Here, we compare levels of particulate polycyclic aromatic compounds (P-PACs, a group of aromatic hydrocarbons containing PAHs and heterocyclic compounds with four or more rings) in air and dermal patch samples from 20 road pavers to the corresponding urinary levels of naphthalene (U-Nap) (two rings), phenanthrene (U-Phe) (three rings), monohydroxylated metabolites of naphthalene (OH-Nap) and phenanthrene (OH-Phe), and 1-hydroxypyrene (OH-Pyr) (four rings), the most widely used biomarker of PAH exposure. For each worker, daily breathing-zone air (n = 55) and dermal patch samples (n = 56) were collected on three consecutive workdays along with postshift, bedtime, and morning urine samples (n = 149). Measured levels of P-PACs and the urinary analytes were used to statistically model exposure-biomarker relationships while controlling for urinary creatinine, smoking status, age, body mass index, and the timing of urine sampling. Levels of OH-Phe in urine collected postshift, at bedtime, and the following morning were all significantly associated with levels of P-PACs in air and dermal patch samples. For U-Nap, U-Phe, and OH-Pyr, both air and dermal patch measurements of P-PACs were significant predictors of postshift urine levels, and dermal patch measurements were significant predictors of bedtime urine levels (all three analytes) and morning urine levels (U-Nap and OH-Pyr only). Significant effects of creatinine concentration were observed for all analytes, and modest effects of smoking status and body mass index were observed for U-Phe and OH-Pyr, respectively. Levels of OH-Nap were not associated with P-PAC measurements in air or dermal patch samples but were significantly affected by smoking status, age, day of sample collection, and urinary creatinine. We conclude that U-Nap, U-Phe, OH-Phe, and OH-Pyr can be used as biomarkers of exposure to particulate asphalt emissions, with OH-Phe being the most promising candidate. Indications that levels of U-Nap, U-Phe, and OH-Pyr were significantly associated with dermal patch measurements well into the evening after a given work shift, combined with the small ratios of within-person variance components to between-person variance components at bedtime, suggest that bedtime measurements may be useful for investigating dermal PAH exposures.

Selenium, folate, and colon cancer.

Selenium is an essential trace element that has been implicated in cancer risk; however, study results have been inconsistent with regard to colon cancer. Our objectives were to 1) investigate the association between selenium and colon cancer, 2) evaluate possible effect measure modifiers, and 3) evaluate potential biases associated with the use of postdiagnostic serum selenium measures. The North Carolina Colon Cancer Study is a large population-based, case-control study of colon cancer in North Carolina between 1996 and 2000 (n = 1,691). Nurses interviewed patients about diet and lifestyle and drew blood specimens, which were used to measure serum selenium. Individuals who had both high serum selenium (> 140 mcg/l) and high reported folate (> 354 mcg/day) had a reduced relative risk of colon cancer [odds ratio (OR) = 0.5, 95% confidence interval (CI) = 0.4-0.8). The risk of colon cancer for those with high selenium and low folate was approximately equal to the risk among those with low selenium and low folate (OR = 1.1, 95% CI = 0.7-1.5) as was the risk for those with low selenium and high folate (OR = 0.9, 95% CI = 0.7-1.2). We did not find evidence of bias due to weight loss, stage at diagnosis, or time from diagnosis to selenium measurement. High levels of serum selenium and reported folate jointly were associated with a substantially reduced risk of colon cancer. Folate status should be taken into account when evaluating the relation between selenium and colon cancer in future studies. Importantly, weight loss, stage at diagnosis, or time from diagnosis to blood draw did not appear to produce strong bias in our study.

Correlates of antioxidant nutrients and oxidative DNA damage differ by race in a cross-sectional study of healthy African American and white adults.

Although antioxidant nutrients and oxidative DNA damage have been associated with carcinogenesis, few studies have investigated the factors that influence antioxidant intake and oxidative DNA damage in racially diverse populations. Demographic, behavioral, and diet-related psychosocial correlates of plasma antioxidant (carotenoids, vitamin C, and vitamin E) concentrations and oxidative DNA damage were examined using data from a cross-sectional study of 147 generally healthy, nonsmoking African American and white adults in North Carolina, aged 20 to 45 years. All participants completed self-administered demographic, diet, and health questionnaires and provided semifasting (> or = 6 hours) blood samples. Multivariate regression analyses were computed separately for each race to determine associations between the potential correlates with plasma antioxidant concentrations and oxidative DNA damage, separately. Our findings suggest appreciable differences by race. Only a few factors (age, supplement use, and several psychosocial factors) were associated with antioxidant concentrations in African Americans, whereas these and additional factors, including physical activity, waist circumference, and passive smoke exposure, were associated with antioxidant concentrations in whites. For oxidative DNA damage, passive smoke exposure was significantly associated with oxidative DNA damage in African Americans, and age and alcohol were significant in whites. In addition, the regression models generally explained more of the variance in plasma antioxidant concentrations and oxidative DNA damage in whites than in African Americans. Considering the salient correlates differed by race, this work has important implications for the design and implementation of future research studies investigating antioxidant nutrients and/or oxidative stress, especially those in racially diverse populations.

Like parent, like child: child food and beverage choices during role playing.

OBJECTIVE: To examine food and beverage choices of preschool-aged children. DESIGN: Semistructured observational study. While pretending to be adults during a role-play scenario, children selected food and beverage items from a miniature grocery store stocked with 73 different products, of which 47 foods and beverages were examined in this analysis. Parents self-reported how frequently they purchased specific grocery items. SETTING: A behavioral laboratory. PARTICIPANTS: One hundred twenty children, aged 2 to 6 years, and 1 parent for each child. Main Outcome Measure Children's total purchases were classified according to the number of healthier and less healthy products they selected as least healthy, somewhat healthy, and most healthy choices. The same categories were used to classify parents' self-reported purchases. RESULTS: Most of the children (70.8%) purchased foods that were categorized as least healthy choices. Only 13 children (10.8%) had shopping baskets consisting of the healthiest choices. On average, children in the group with the least healthy choices purchased the same number of healthier and less healthy products, whereas children in the group with most healthy choices purchased 5 healthier products for each less healthy product selected. The healthfulness of children's total purchases were significantly (P = .02) predicted by their parents' purchasing categorization. CONCLUSIONS: When presented with a wide array of food products, young children chose combinations of healthier and less healthy foods and beverages. The data suggest that children begin to assimilate and mimic their parents' food choices at a very young age, even before they are able to fully appreciate the implications of these choices.

Quantifying the relative importance of predictors in multiple linear regression analyses for public health studies.

Multiple linear regression analysis is widely used in many scientific fields, including public health, to evaluate how an outcome or response variable is related to a set of predictors. As a result, researchers often need to assess "relative importance" of a predictor by comparing the contributions made by other individual predictors in a particular regression model. Hence, development of valid statistical methods to estimate the relative importance of a set of predictors is of great interest. In this research, the authors considered the relative importance of a predictor when defined by that portion of the squared multiple correlation explained by the contribution of each predictor in the final model of interest. Here, a number of suggested relative importance indices motivated by this definition are reviewed, including the squared zero-order correlation, squared semipartial correlation, Product Measure (i.e., Pratt's Index), General Dominance Index, and Johnson's Relative Weight. The authors compared these indices using data sets from an occupational health study in which human inhalation exposure to styrene was measured and from a laboratory animal study on risk factors for atherosclerosis, and statistical properties using bootstrap methods were examined. The analysis suggests that the General Dominance Index and Johnson's Relative Weight are preferred methods for quantifying the relative importance of predictors in a multiple linear regression model. Johnson's Relative Weight involves significantly less computational burden than the General Dominance Index when the number of predictors in the final model is large.

Physical and sexual violence among North Carolina women: associations with physical health, mental health, and functional impairment.

OBJECTIVE: This study examines links between women's experiences of violence during adulthood (including physical and sexual violence) and women's physical health, mental health, and functional status. METHODS: Data were analyzed from a representative sample of 9,830 North Carolina women surveyed by the North Carolina Behavioral Risk Factor Surveillance System (BRFSS). RESULTS: One-quarter of the women experienced violence as adults, with current or ex-partners being the most common perpetrators. Logistic regression analyses that controlled for the sociodemographic characteristics of the women found that women who experienced violence were significantly more likely than other women to have poor physical health, poor mental health, and functional limitations. Moreover, these negative health outcomes were most prevalent among the women who experienced a combination of both physical and sexual violence. CONCLUSIONS: These findings underscore the need for trauma-informed women's health services and policies.