Quantitative Exposures Research Articles

The statistics of time varying cross-sectional information coefficients

The information coefficient (IC), defined as the correlation coefficient between a stock return and its factor exposures predictor variables, is one of the most commonly used statistics in quantitative financial analysis. In this paper, we establish consistency and asymptotic normality of the time series average of cross-sectional sample ICs when the true underlying ICs between the risk-adjusted residual return and the standardized factor exposures are time varying. We use those results to show that the time series average of the cross-sectional sample ICs divided by its sample standard deviation converges to the ex ante expected portfolio information ratio (IR) as derived in Ding and Martin (2017). A simulation study based on a true factor model shows that the finite sample results are strikingly close to what the theory suggests. We also conduct empirical simulations using actual stock returns and quantitative factor exposures, and we find that the logarithm of the estimated IR can be explained very well by a function of the IC mean, the IC standard deviation, and the sample size in exactly the same way as predicted by our theory built on a linear factor model with time varying ICs.

Levels of persistent organic pollutants in breast milk samples representing Finnish and Danish boys with and without hypospadias

Hypospadias is a congenital malformation of penile urethra with unknown etiology in most cases. Persistent organic pollutant (POP) exposure may disrupt endocrine function during a critical window of development of male genitalia. In animal studies, POPs have been associated with male reproductive disorders, including hypospadias, but only few studies have assessed this relationship in humans. The aim of this study is to investigate the association between hypospadias and POP concentration levels in breast milk, as a proxy for prenatal exposure. This is a nested case-control study of Danish and Finnish mother-son pairs. Maternal breast milk samples were collected between 1997 and 2002, and they represent infant boys born with hypospadias [n = 33 (n = 22 Danish and n = 11 Finnish)] and their 1:1 matched controls. Breast milk samples were analyzed for six classes of POPs [including dioxins, polychlorinated biphenyls, flame retardants and perfluorinated alkylated substances (PFAS)]. We estimated odds ratios (ORs) and 95% confidence intervals (CI) for each chemical class using conditional logistic regression. In addition, a composite exposure score system was used to explore the effect of a POP mixture (four chemical classes): The composite score was categorized as low, moderate, or high exposure, and differences between cases and controls were tested with conditional logistic regression. No statistically significant associations were observed between the sums of the chemical classes and hypospadias in either country. The composite score was unable to detect differences in the risk of hypospadias between the tertiles of POP exposure. Levels of PFAS were significantly higher in Danish than in Finnish breast milk samples. This small study does not provide evidence for an association between hypospadias and exposure to POPs but adds information on quantitative exposures. Further development of multi-exposure models is needed for assessing the potential mixture effect associated with multiple chemical exposures.

Exploration of occupations as risk factors for lung cancer in hierarchical and penalization penalized regression models accounting for multiple known exposures

Background and aim: Traditional occupational epidemiological analysis typically emphasize on disease risks associated with specific single exposures or single occupations. We used hierarchical and regularization models to explore occupational risks at the job-title level associated with lung cancer while accounting for exposures to multiple known lung carcinogens. Methods: We pooled lung cancer case-control study subjects from 14 European and Canadian studies. Associations between ever employment in 1,288 five-digit ISCO-68 occupations and lung cancer were screened using two approaches: Bayesian hierarchical and LASSO penalized regressions. Both modeling approaches account for age, smoking, sex, study, and fully quantitative exposures to six known occupational lung carcinogens: asbestos, chromium, diesel engine exhaust, nickel, PAHs, and silica. False positive findings in the Bayesian hierarchical and LASSO models were controlled using horseshoe priors and stability selection, respectively. Lung cancer odds ratios for selected occupations were estimated using unconditional logistic regression model with identical covariates. Results: Our final analyses included 16,901 lung cancer cases and 20,965 controls. There were notable overlaps in jobs selected by the hierarchical and penalization modeling approaches. Occupations selected by both modeling approaches with positive associations with lung cancer after controlling for known exposures include carpenters (OR: 1.77; 95CI: 1.36, 2.33) and construction bricklayers (OR: 1.40; 95CI: 1.17, 1.67). Similarly selected occupations but with negative associations with lung cancer include secondary school teachers (OR: 0.36; 95CI: 0.23, 0.55), production managers (OR: 0.65; 95CI: 0.53, 0.79), and government executive officials (OR: 0.7; 95% CI: 0.61, 0.79). Conclusion: We demonstrated viable agnostic approaches in identifying employment risk factors for lung cancer. Future work involves investigations of factors that contribute to the observed elevated or lower cancer risks, which may involve gaps in current exposure assessment and modeling framework, and/or exposures to other lung carcinogens yet to be elucidated.

Variance estimators for weighted and stratified linear dose-response function estimators using generalized propensity score.

Propensity score methods are widely used in observational studies for evaluating marginal treatment effects. The generalized propensity score (GPS) is an extension of the propensity score framework, historically developed in the case of binary exposures, for use with quantitative or continuous exposures. In this paper, we proposed variance estimators for treatment effect estimators on continuous outcomes. Dose-response functions (DRFs) were estimated through weighting on the inverse of the GPS, or using stratification. Variance estimators were evaluated using Monte Carlo simulations. Despite the use of stabilized weights, the variability of the weighted estimator of the DRF was particularly high, and none of the variance estimators (a bootstrap-based estimator, a closed-form estimator especially developed to take into account the estimation step of the GPS, and a sandwich estimator) were able to adequately capture this variability, resulting in coverages below the nominal value, particularly when the proportion of the variation in the quantitative exposure explained by the covariates was large. The stratified estimator was more stable, and variance estimators (a bootstrap-based estimator, a pooled linearized estimator, and a pooled model-based estimator) more efficient at capturing the empirical variability of the parameters of the DRF. The pooled variance estimators tended to overestimate the variance, whereas the bootstrap estimator, which intrinsically takes into account the estimation step of the GPS, resulted in correct variance estimations and coverage rates. These methods were applied to a real data set with the aim of assessing the effect of maternal body mass index on newborn birthweight.

Metabolite V, an epoxide species is a minor circulating metabolite in humans following a single oral dose of deflazacort.

Deflazacort (Emflaza) was approved in the United States in 2017 for the treatment of the Duchenne muscular dystrophy in patients aged 2 years and older. Several deflazacort metabolites were isolated and identified from rats, dogs, monkeys, and humans. Among them, 1ß,2ß‐epoxy‐3ß‐hydroxy‐21‐desacetyl deflazacort, referred to as Metabolite V, was reported to be one of the major circulating metabolites in humans. However, its quantitative distribution in plasma was not fully characterized. The objective of this study was to determine deflazacort plasma pharmacokinetics, metabolite profiles and their quantitative exposures in humans following a single oral dose. Six healthy male subjects were each administered a single oral dose of 60 mg [14C]‐deflazacort. Plasma and urine were collected and deflazacort metabolites in plasma were quantified by high performance liquid chromatography radio‐profiling followed by liquid chromatography‐mass spectrometry characterization. Metabolite V was isolated from urine and its structure was further confirmed by nuclear magnetic resonance analysis. These analyses demonstrated that deflazacort was not detectable in plasma; of the eight circulating deflazacort metabolites identified or characterized, the pharmacologically active metabolite 21‐desacetyl deflazacort and inactive metabolite 6ß‐hydroxy‐21‐desacetyl deflazacort accounted for 25.0% and 32.9% of the 0‐24 hours plasma total radioactivity, respectively, while Metabolite V, an epoxide species, was a minor circulating metabolite, representing only about 4.7% of the total plasma radioactivity.

Assessing the performance of the generalized propensity score for estimating the effect of quantitative or continuous exposures on survival or time-to-event outcomes.

Propensity score methods are frequently used to estimate the effects of interventions using observational data. The propensity score was originally developed for use with binary exposures. The generalized propensity score (GPS) is an extension of the propensity score for use with quantitative or continuous exposures (e.g. pack-years of cigarettes smoked, dose of medication, or years of education). We describe how the GPS can be used to estimate the effect of continuous exposures on survival or time-to-event outcomes. To do so we modified the concept of the dose–response function for use with time-to-event outcomes. We used Monte Carlo simulations to examine the performance of different methods of using the GPS to estimate the effect of quantitative exposures on survival or time-to-event outcomes. We examined covariate adjustment using the GPS and weighting using weights based on the inverse of the GPS. The use of methods based on the GPS was compared with the use of conventional G-computation and weighted G-computation. Conventional G-computation resulted in estimates of the dose–response function that displayed the lowest bias and the lowest variability. Amongst the two GPS-based methods, covariate adjustment using the GPS tended to have the better performance. We illustrate the application of these methods by estimating the effect of average neighbourhood income on the probability of survival following hospitalization for an acute myocardial infarction.

1013 Novel Quantitative EEG Exposures and Risk of Incident MCI and Dementia in Older Women

1013 Novel Quantitative EEG Exposures and Risk of Incident MCI and Dementia in Older Women

Assessing the performance of the generalized propensity score for estimating the effect of quantitative or continuous exposures on binary outcomes

Propensity score methods are increasingly being used to estimate the effects of treatments and exposures when using observational data. The propensity score was initially developed for use with binary exposures. The generalized propensity score (GPS) is an extension of the propensity score for use with quantitative or continuous exposures (eg, dose or quantity of medication, income, or years of education). We used Monte Carlo simulations to examine the performance of different methods of using the GPS to estimate the effect of continuous exposures on binary outcomes. We examined covariate adjustment using the GPS and weighting using weights based on the inverse of the GPS. We examined both the use of ordinary least squares to estimate the propensity function and the use of the covariate balancing propensity score algorithm. The use of methods based on the GPS was compared with the use of G‐computation. All methods resulted in essentially unbiased estimation of the population dose‐response function. However, GPS‐based weighting tended to result in estimates that displayed greater variability and had higher mean squared error when the magnitude of confounding was strong. Of the methods based on the GPS, covariate adjustment using the GPS tended to result in estimates with lower variability and mean squared error when the magnitude of confounding was strong. We illustrate the application of these methods by estimating the effect of average neighborhood income on the probability of death within 1 year of hospitalization for an acute myocardial infarction.

Assessing covariate balance when using the generalized propensity score with quantitative or continuous exposures.

Propensity score methods are increasingly being used to estimate the effects of treatments and exposures when using observational data. The propensity score was initially developed for use with binary exposures (e.g., active treatment vs. control). The generalized propensity score is an extension of the propensity score for use with quantitative exposures (e.g., dose or quantity of medication, income, years of education). A crucial component of any propensity score analysis is that of balance assessment. This entails assessing the degree to which conditioning on the propensity score (via matching, weighting, or stratification) has balanced measured baseline covariates between exposure groups. Methods for balance assessment have been well described and are frequently implemented when using the propensity score with binary exposures. However, there is a paucity of information on how to assess baseline covariate balance when using the generalized propensity score. We describe how methods based on the standardized difference can be adapted for use with quantitative exposures when using the generalized propensity score. We also describe a method based on assessing the correlation between the quantitative exposure and each covariate in the sample when weighted using generalized propensity score -based weights. We conducted a series of Monte Carlo simulations to evaluate the performance of these methods. We also compared two different methods of estimating the generalized propensity score: ordinary least squared regression and the covariate balancing propensity score method. We illustrate the application of these methods using data on patients hospitalized with a heart attack with the quantitative exposure being creatinine level.

Breast Cancer Incidence and Exposure to Metalworking Fluid in a Cohort of Female Autoworkers.

Breast cancer is the leading cancer diagnosed among women, and environmental studies have produced few leads on modifiable risk factors for breast cancer. Following an Institute of Medicine recommendation for occupational studies of women highly exposed to potential breast cancer risk factors, we took advantage of an existing cohort of 4,503 female autoworkers in Michigan exposed to metalworking fluid (MWF), complex mixtures of oils and chemicals widely used in metal manufacturing worldwide. Cox proportional hazards models were fit to estimate hazard ratios for incident breast cancer (follow-up, 1985-2013) and cumulative exposure (20-year lag) to straight mineral oils (a known human carcinogen) and water-based soluble and synthetic MWF. Because the state cancer registry began decades after the cohort was defined, we restricted our analyses to subcohorts of women hired closer to the start of follow-up. Among those hired after 1969, the hazard ratio associated with a 1 interquartile-range increase in straight MWF exposure was 1.13 (95% confidence interval: 1.03, 1.23). In separate analyses of premenopausal breast cancer, defined by age at diagnosis, the hazard ratio was elevated for exposure to synthetic MWF (chemical lubricants with no oil content), possibly suggesting a different mechanism in the younger women with breast cancer. This study adds to the limited literature regarding quantitative chemical exposures and breast cancer risk.

Evaluation of Oil Fate and Exposure from a Deep Water Blowout With and Without Subsea Dispersant Injection Treatment as Well as Traditional Response Activities

ABSTRACT The goals of subsea dispersant injection (SSDI) into a deep water oil and gas blowout are to increase effectiveness of dispersant treatment over that achievable at the water surface; decrease the volume of oil that surfaces; reduce human and wildlife exposure to volatile organic compounds (VOCs); disperse the oil over a large water volume at depth; enhance biodegradation; and reduce surface, nearshore and shoreline exposure to floating and surface-water entrained/dissolved oil. Potential tradeoffs include increased water column and benthic resource exposures to oil at depth. In order to better understand the implications of SSDI use, we modeled a hypothetical blowout in the northern Gulf of Mexico to predict oil fate and compare the environmental exposure for no intervention to various response options (i.e., mechanical recovery, in-situ burning (ISB), surface dispersant application, and SSDI). Probabilistic modeling was used to evaluate the influence of variable metocean conditions (i.e., wind, currents, temperature). The results showed that even with a substantial capacity of equipment applied, mechanical and ISB removed only a small fraction of the oil that would otherwise be floating or evaporate. Compared to cases without use of SSDI, SSDI reduced the size of oil droplets by an order of magnitude, substantially decreased the amount of oil on the water surface and on the shoreline, increased dissolution and degradation rates of hydrocarbons at depth, increased weathering rate of rising oil such that floating oil contained much lower content of soluble and semi-soluble hydrocarbons, decreased surface water concentrations of dissolved hydrocarbons, and decreased VOC emissions to the atmosphere and, therefore, reduced human and wildlife exposures to VOCs. The tradeoff was that with SSDI there was greater exposure to hydrocarbons in deep water. However, densities of biota are much lower in deep water than near the water surface, where sensitive early life history stages of fish and invertebrates are most abundant. This approach provides decision makers with quantitative environmental exposures with which they may evaluate risk tradeoffs regarding appropriate response strategies for mitigating impacts from oil and gas released during a deep water blowout.

Assessing the reporting of categorised quantitative variables in observational epidemiological studies

BackgroundOne aspect to consider when reporting results of observational studies in epidemiology is how quantitative risk factors are analysed. The STROBE (Strengthening the Reporting of Observational Studies in Epidemiology) guidelines recommend that researchers describe how they handle quantitative variables when analysing data. For categorised quantitative variables, the authors are required to provide reasons and justifications informing their practice. We investigated and assessed the practices and reporting of categorised quantitative variables in epidemiology.MethodsThe assessment was based on five medical journals that publish epidemiological research. Observational studies published between April and June 2015 and investigating the relationships between quantitative exposures (or risk factors) and the outcomes were considered for assessment. A standard form was used to collect the data, and the reporting patterns amongst eligible studies were quantified and described.ResultsOut of 61 articles assessed for eligibility, 23 observational studies were included in the assessment. Categorisation of quantitative exposures occurred in 61% of these studies and reasons informing the practice were rarely provided. Only one article explained the choice of categorisation in the analysis. Transformation of quantitative exposures into four or five groups was common and dominant amongst studies using equally spaced categories. Dichotomisation was not popular; the practice featured in one article. Overall, the majority (86%) of the studies preferred ordered or arbitrary group categories. Other criterions used to decide categorical boundaries were based on established guidelines such as consensus statements and WHO standards.ConclusionCategorisation of continuous variables remains a dominant practice in epidemiological studies. The reasons informing the practice of categorisation within published work are limited and remain unknown in most articles. The existing STROBE guidelines could provide stronger recommendations on reporting quantitative risk factors in epidemiology.

Occupational exposure to solvents and bladder cancer: A population-based case control study in Nordic countries.

The objective of the study was to assess the relationship between exposure to selected solvents and the risk of bladder cancer. This study is based on the Nordic Occupational Cancer (NOCCA) database and comprises 113,343 cases of bladder cancer diagnosed in Finland, Iceland, Norway and Sweden between 1961 and 2005 and 566,715 population controls matched according to country, sex and birth year. Census-based occupational titles of the cases and controls were linked with the job exposure matrix created by the NOCCA project to estimate quantitative cumulative occupational exposures. A conditional logistic regression model was used to estimate hazard ratios (HRs) and their 95% confidence intervals (95% CI). Increased risks were observed for trichloroethylene (HR 1.23, 95% CI 1.12-1.40), toluene (HR 1.20, 95% CI 1.00-1.38), benzene (HR 1.16, 95% CI 1.04-1.31), aromatic hydrocarbon solvents (HR 1.10, 95% CI 0.94-1.30) and aliphatic and alicyclic hydrocarbon solvents (HR 1.08, 95% CI 1.00-1.23) at high exposure level versus no exposure. The highest excess for perchloroethylene was observed at medium exposure level (HR 1.12, 95% CI 1.02-1.23). The study provides evidence of an association of occupational exposure to trichloroethylene, perchloroethylene, aromatic hydrocarbon solvents, benzene and toluene and the risk of bladder cancer.

0122 Approaches to developing exposure estimates that reflect temporal trends in total particulate matter in aluminium smelters

ObjectivesTo evaluate different approaches for including time trends in quantitative exposures to total particulate matter (TPM) in an aluminium smelter for use in epidemiologic evaluation of incident heart disease, we...

CARDIOVASCULAR MORTALITY AND OCCUPATIONAL EXPOSURE IN AN ALUMINUM COHORT

Background and aims: Increased risk of cardiovascular mortality related to combustion-generated particulate matter (PM) in air pollution has become a major public health concern. Although occupational exposures to PM can be orders of magnitude higher than ambient levels, evidence regarding cardiovascular disease risk associated with workplace exposure to PM is scant. Methods: We examined cardiovascular mortality from 1996 to 2008 in a cohort of over 11,700 aluminum workers which has been under observation for over a decade. Individual level exposure metrics for total PM and PM2.5 are being developed for the cohort based on job histories and personal sampling results in smelters, refining and fabrication facilities. Results: Initial analyses using broad classifications of plant type (smelter vs other) show no increased risk of all-cause mortality among those who worked in a smelter. Thirty percent of the deaths were due to cardiovascular disease (CVD). Stratifying by active versus inactive person-time, we found a 14% increased risk for death from CVD (N = 78) and a 28% increased risk for death from ischemic heart disease (N = 55) for active workers who worked at smelters, although all confidence intervals include the null. Conclusions: Higher peak and mean PM exposures occurred at smelting facilities compared to refining and fabrication facilities, lending support to the hypothesis that there is an exposure-response association between quantitative exposures to PM and CVD mortality.

Quantitative exposure matrix for asphalt fume, total particulate matter, and respirable crystalline silica among roofing and asphalt manufacturing workers

This paper summarizes available data on worker exposures to asphalt fume (soluble fraction), total particulate matter, and respirable crystalline silica (quartz) [hereinafter RCS] over a 30-year period in Owens Corning’s asphalt production and roofing manufacturing plants. For the period 1977 through 2006, the air-monitoring database contains more than 1,400 personal samples for asphalt fume (soluble fraction), 2,400 personal samples for total particulate, and 1,300 personal samples for RCS. Unique process-job categories were identified for the asphalt production and roofing shingle manufacturing plants. Quantitative exposures were tabulated by agent, process-job, and calendar period to form an exposure matrix for use in subsequent epidemiologic studies of the respiratory health of these workers. Analysis of time trends in exposure data shows substantial and statistically significant exposure reductions for asphalt fume (soluble fraction), total particulate matter, and respirable crystalline silica at Owens Corning plants. Cumulative distribution plots for the most recent sampling period (2001–2006) show that 95% of the asphalt fume (soluble fraction) measurements were less than 0.25 mg/m3; 95% of the total particulate measurements were less than 2.2 mg/m3; and 95% of the RCS measurements were less than 0.05 mg/m3. Several recommendations are offered to improve the design of future monitoring efforts.

Quantitative Crystalline Silica Exposure Assessment for a Historical Cohort Epidemiologic Study in the German Porcelain Industry

A time-dependent quantitative exposure assessment of silica exposure among nearly 18,000 German porcelain workers was conducted. Results will be used to evaluate exposure-response disease risks. Over 8000 historical industrial hygiene (IH) measurements with original sampling and analysis protocols from 1954–2006 were obtained from the German Berufs- genossenschaft der keramischen-und Glas-Industrie (BGGK) and used to construct a job exposure matrix (JEM). Early measurements from different devices were converted to modern gravimetric equivalent values. Conversion factors were derived from parallel historical measurements and new side-by-side measurements using historical and modern devices in laboratory dust tunnels and active workplace locations. Exposure values were summarized and smoothed using LOESS regression; estimates for early years were derived using backward extrapolation techniques. Employee work histories were merged with JEM values to determine cumulative crystalline silica exposures for cohort members. Average silica concentrations were derived for six primary similar exposure groups (SEGs) for 1938–2006. Over 40% of the cohort accumulated <0.5 mg; just over one-third accumulated >1 mg/m3-years. Nearly 5000 workers had cumulative crystalline silica estimates >1.5 mg/m3-years. Similar numbers of men and women fell into each cumulative exposure category, except for 1113 women and 1567 men in the highest category. Over half of those hired before 1960 accumulated >3 mg/m3-years crystalline silica compared with 4.9% of those hired after 1960. Among those ever working in the materials preparation area, half accumulated >3 mg/m3-year compared with 12% of those never working in this area. Quantitative respirable silica exposures were estimated for each member of this cohort, including employment periods for which sampling used now obsolete technologies. Although individual cumulative exposure estimates ranged from background to about 40 mg/m3-years, many of these estimates reflect long-term exposures near modern exposure limit values, allowing direct evaluation of lung cancer and silicosis risks near these limits without extrapolation. This quantitative exposure assessment is the largest to date in the porcelain industry.

Quantitative Exposure to Metalworking Fluids and Bladder Cancer Incidence in a Cohort of Autoworkers

Occupations with mineral oil exposure have been associated with bladder cancer in population-based case-control studies. The authors report results from the first cohort study to examine bladder cancer incidence in relation to quantitative exposures to metalworking fluids (MWFs), based on 21,999 male Michigan automotive workers, followed from 1985 through 2004. Cox regression was used to estimate hazard ratios based on categorical exposure variables for straight, soluble, and synthetic MWFs, as well as duration of exposure to ethanolamines and nitrosamines. Penalized splines were also fit to estimate the functional form of the exposure-response relation. Increased bladder cancer risk was associated with straight MWFs but not with any other exposure. The hazard ratio increased with cumulative exposure to a maximum of 2-fold observed at 75 mg/m(3)-year straight MWF exposure (lagged 20 years). Calendar time windows relevant to polycyclic aromatic hydrocarbon exposure were examined but could not be distinguished from the lagged (10-, 20-year) metrics. No association was observed between any exposure and incident lung cancer, suggesting that smoking is unlikely to confound the associations observed here. The quantitative relation with straight MWFs strengthens the evidence for mineral oils as a bladder carcinogen.

Occupational exposures to magnetic fields and neurodegenerative disease risks

Occupational exposures to magnetic fields and neurodegenerative disease risks

Female Survivors of Childhood Cancer: Preterm Birth and Low Birth Weight Among Their Children

Improved survival after childhood cancer raises concerns over the possible long-term reproductive health effects of cancer therapies. We investigated whether children of female childhood cancer survivors are at elevated risk of being born preterm or exhibiting restricted fetal growth and evaluated the associations of different cancer treatments on these outcomes. Using data from the Childhood Cancer Survivor Study, a large multicenter cohort of childhood cancer survivors, we studied the singleton live births of female cohort members from 1968 to 2002. Included were 2201 children of 1264 survivors and 1175 children of a comparison group of 601 female siblings. Data from medical records were used to determine cumulative prepregnancy exposures to chemotherapy and radiotherapy. Logistic regression was used to estimate odds ratios (ORs) for the association between quantitative therapy exposures and preterm (<37 weeks) birth, low birth weight (<2.5 kg), and small-for-gestational-age (SGA) (lowest 10th percentile) births. All statistical tests were two-sided. Survivors' children were more likely to be born preterm than the siblings' children (21.1% versus 12.6%; OR = 1.9, 95% confidence interval [CI] = 1.4 to 2.4; P<.001). Compared with the children of survivors who did not receive any radiotherapy, the children of survivors treated with high-dose radiotherapy to the uterus (>500 cGy) had increased risks of being born preterm (50.0% versus 19.6%; OR = 3.5, 95% CI = 1.5 to 8.0; P = .003), low birth weight (36.2% versus 7.6%; OR = 6.8, 95% CI = 2.1 to 22.2; P = .001), and SGA (18.2% versus 7.8%; OR = 4.0, 95% CI = 1.6 to 9.8; P = .003). Increased risks were also apparent at lower uterine radiotherapy doses (starting at 50 cGy for preterm birth and at 250 cGy for low birth weight). Late effects of treatment for female childhood cancer patients may include restricted fetal growth and early births among their offspring, with risks concentrated among women who receive pelvic irradiation.