Residential indoor PM2.5 in wood stove homes: follow-up of the Libby changeout program.

UNLABELLED: In 2005 through 2008, a small rural mountain valley community engaged in a woodstove changeout program to address concerns of poor ambient air quality. During this program, we assessed changes to indoor air quality before and after the introduction of a new, lower emission woodstove. We previously reported a >70% reduction in indoor PM(2.5) concentrations in homes following the installation of a new Environmental Protection Agency's-certified stove within the home. We report here on follow-up of the experiences in these and other homes over three winters of sample collection. In 21 homes, we compared pre-changeout PM(2.5) concentrations [mean (s.d.) = 45.0 (33.0) µg/m(3)] to multiple post-changeout measures of PM(2.5) concentrations using a DustTrak. The mean reduction (and 95% confidence interval) from pre-changeout to post-changeout was -18.5 µg/m(3) (-31.9, -5.2), adjusting for ambient PM(2.5) , ambient temperature, and other factors. Findings across homes and across years were highly variable, and a subset of homes did not experience a reduction in PM(2.5) following changeout. Reductions were also observed for organic carbon, elemental carbon, and levoglucosan, but increases were observed for dehydroabietic acid and abietic acid. Despite overall improvements in indoor air quality, the varied response across homes may be due to factors other than the introduction of a new woodstove. PRACTICAL IMPLICATIONS: Biomass combustion is a common source of ambient PM(2.5) in many cold-climate communities. The replacement of older model woodstoves with newer technology woodstoves is a potential intervention strategy to improve air quality in these communities. In addition to ambient air, woodstove changeouts should improve residential indoor air quality. We present results from a multi-winter study to evaluate the efficacy of woodstove changeouts on improving indoor air quality. Reductions in indoor PM(2.5) were evident, but this observation was not consistent across all homes. These findings suggest that other factors beyond the introduction of an improved wood burning device are relevant to improving indoor air quality in wood burning homes.

Unequal exchange at the Charcot-Marie-Tooth disease type 1A recombination hot-spot is not elevated above the genome average rate.

An increasing number of human diseases and syndromes are being found to result from micro-duplications or microdeletions arising from meiotic recombination between homologous repeats on the same chromosome. The first microduplication syndrome delineated, Charcot-Marie-Tooth disease type 1A (CMT1A), results from unequal crossing over between two >98% identical 24 kb repeats (CMT1A-REPs) on chromosome 17. In addition to its medical significance, the CMT1A region has features that make it a unique resource for detailed analysis of human unequal recombination. Previous studies of CMT1A patients showed that the majority of unequal crossovers occurred within a small region (<1 kb) of the REPs suggesting the presence of a recombination hot-spot. We directly measured the frequency of unequal recombination in the hot-spot region using sperm from four normal individuals. Surprisingly, unequal recombination between the REPs occurs at a rate no greater than the average rate for the male genome (approximately 1 cM/Mb) and is the same as that expected for equally aligned REPs. This conclusion extends to humans the findings in yeast that recombination between repeated sequences far apart on the same chromosome may occur at similar frequencies to allelic recombination. Finally, the CMT1A hot-spot stands in sharp contrast to the human MS32 mini-satellite-associated hot-spot that exhibits highly enhanced recombination initiation in addition to positional specificity. One possibility is that the CMT1A hot-spot may consist of a region with genome average recombination potential embedded within a recombination cold-spot.

Statistical methods for epidemiologic studies of the health effects of air pollution.

We describe two statistical designs that can provide efficient estimates of the health effects of exposure to air pollutants in epidemiologic studies. We also evaluate the effects of measurement error in exposure assessment on the accuracy of estimated health effects. The bidirectional case-crossover design is a variant of a method proposed by Maclure (1991). Our version of the method takes advantage of the fact that in epidemiologic studies involving environmental exposure, accurate information about past exposure is more readily available, and that levels of exposure are generally unaffected by the response of the subject. It differs from other case-crossover methods in that control information is assessed both before and after failure, thus avoiding confounding due to time trends in exposure. The multilevel analytic design provides a method of combining estimates of health effects made on the individual level with those made at the group level. It has great potential value in situations where variations in exposure within groups may not be great enough to provide adequate power to detect health effects, as is often the case in air pollution studies where exposure levels are similar within a geographic community. Measurement errors in exposure assessment can have substantial impact on the accuracy of estimated health effects. When the microenvironmental approach is used to estimate exposure, a standard error of 30% in estimating indoor/outdoor ratios can increase the standard error of a relative risk estimate by 50%, and introduce bias as well. Similar results hold when exposure is estimated with personal samplers. When the microenvironmental approach is used, errors in estimating indoor/outdoor ratios have more influence on the accuracy of risk estimation than do errors in estimating the time spent in microenvironments.

A study of twelve Southern California communities with differing levels and types of air pollution. II. Effects on pulmonary function.

To study the possible chronic respiratory effects of air pollutants, we designed and initiated a 10-yr prospective study of Southern California public schoolchildren living in 12 communities with different levels and profiles of air pollution. The design of the study, exposure assessment methods, and survey methods and results related to respiratory symptoms and conditions are described in the accompanying paper. Pulmonary function tests were completed on 3,293 subjects. We evaluated cross-sectionally the effects of air pollution exposures based on data collected in 1986-1990 by existing monitoring stations and data collected by our study team in 1994. Expected relationships were seen between demographic, physical, and other environmental factors and pulmonary function values. When the data were stratified by sex, an association was seen between pollution levels and lower pulmonary function in female subjects, with the associations being stronger for the 1994 exposure data than the 1986-1990 data. After adjustment, PM10, PM2.5, and NO2 were each significantly associated with lower FVC, FEV1, and maximal midexpiratory flow (MMEF); acid vapor with lower FVC, FEV1, peak expiratory flow rate (PEFR), and MMEF; and O3 with lower PEFR and MMEF. Effects were generally larger in those girls spending more time outdoors. Stepwise regression of adjusted pulmonary function values for girls in the 12 communities showed that NO2 was most strongly associated with lower FVC (r = -0.74, p < 0.01), PM2.5 with FEV1 (r = -0.72, p < 0.01), O3 with PEFR (r = -0.75, p < 0.005), and PM2.5 with MMEF (r = -0.80, p < 0.005). There was a statistically significant association between ozone exposure and decreased FVC and FEV1 in girls with asthma. For boys, significant associations were seen between peak O3 exposures and lower FVC and FEV1, but only in those spending more time outdoors. These findings underline the importance of follow-up of this cohort.

A study of twelve Southern California communities with differing levels and types of air pollution. I. Prevalence of respiratory morbidity.

To study possible chronic respiratory effects of air pollutants, we initiated a 10-yr prospective cohort study of Southern California children, with a study design focused on four pollutants: ozone, particulate matter, acids, and nitrogen dioxide (NO2). Twelve demographically similar communities were selected on the basis of historic monitoring information to represent extremes of exposure to one or more pollutants. In each community, about 150 public school students in grade 4, 75 in grade 7, and 75 in grade 10 were enrolled through their classrooms. Informed consent and written responses to surveys about students' lifetime residential histories, historic and current health status, residential characteristics, and physical activity were obtained with the help of the parents. In the first testing season, 3,676 students returned questionnaires. We confirmed associations previously reported between respiratory morbidity prevalence and the presence of personal, demographic, and residential risk factors. Rates of respiratory illness were higher for males, those living in houses with pets, pests, mildew, and water damage, those whose parents had asthma, and those living in houses with smokers. Wheeze prevalence was positively associated with levels of both acid (odds ratio [OR] = 1.45; 95% confidence interval [CI], 1.14-1.83) and NO2 (OR = 1.54; 95% CI, 1.08-2.19) in boys. We conclude, based on this cross-sectional assessment of questionnaire responses, that current levels of ambient air pollution in Southern California may be associated with effects on schoolchildren's respiratory morbidity as assessed by questionnaire.

Bidirectional case-crossover designs for exposures with time trends.

In the case-crossover design (Maclure, 1991, American Journal of Epidemiology 133, 144-153), only cases are sampled, and risk estimates are based on within-subject comparisons of exposures at failure times with exposures at times prior to failure, using matched case-control methods. While the design provides considerable advantages, unidirectional retrospective control sampling (selecting control times only prior to failure) can cause risk estimates to be confounded by time trends in exposure. However, when subsequent exposures are not influenced by failures, as in studies of environmental exposures such as air pollutants, it is possible to determine at times postfailure what a subject's level of exposure would have been had the subject not failed. We describe a bidirectional case-crossover design in which exposures at failure are compared with exposures both before and after failure. Simulation analyses show that relative risk estimates are resistant to confounding by time trend. We also extend the method to studies involving multiple failure times.

Acute effects of ambient ozone on asthmatic, wheezy, and healthy children.

Southern California children (10 to 12 years old) participated in a two-season study to assess the potential acute respiratory effects of ambient ozone (O3). Asthmatic (n = 49), wheezy (n = 53), and healthy (n = 93) children completed a four-day (Friday through Monday) study protocol, once in spring and again in summer, that included the use of daily activity and symptom diaries, heart rate recording devices, personal O3 samplers, and maximal effort spirometry several times per day. Data from regional monitoring stations were used to establish ambient hourly O3 concentrations. Analyses revealed that the children spent more time outdoors and were more physically active in the spring. Girls spent less time outdoors and were less physically active than boys. Personal O3 samplers correlated poorly with, and generally gave lower readings than, outdoor ambient monitors. Higher personal O3 exposures were associated generally with increased inhaler use, more outdoor time, and more physical activity. Children with asthma spent more time outdoors and were more active in the spring on high-O3 days (measured by personal sampler), and had the most trouble breathing, the most wheezing, and the most inhaler use on these days. Activity pattern data suggested that children with asthma protected themselves by being less physically active outdoors during the summer on high-O3 days. Wheezy children had the most trouble breathing during the summer on low-O3 days (measured by personal sampler). Observed relationships between O3 and pulmonary function were erratic and difficult to reconcile with existing knowledge about the acute respiratory effects of air pollution. We conclude that although asthmatic and wheezy children behave differently from their healthy peers with regard to symptoms and patterns of activity when challenged by ambient ozone, the nature of these changes remains inconsistent and ill-defined.

Genetic polymorphism of CYP2D6 and lung cancer risk in African-Americans and Caucasians in Los Angeles County.

The well described genetic polymorphism of the CYP2D6 gene influences response to a wide variety of therapeutic agents metabolized by the CYP2D6 enzyme product. CYP2D6 also appears to play a role, along with other cytochrome P450 enzymes, in the metabolic activation of the tobacco specific nitrosamine, NNK, as well as metabolism of nicotine to cotinine. While impaired activity of CYP2D6 was strongly protective against lung cancer in some studies, primarily based on phenotyping, the literature is conflicting. The molecular basis of CYP2D6 deficiency is now well understood, enabling the use of genotyping to classify individuals. We therefore examined whether lung cancer risk is reduced by the presence of four CYP2D6 alleles associated with impaired activity due to an inactivating mutation--CYP2D6*4, CYP2D6*3, CYP2D6*5 and CYP2D6*16--among 341 incident cases of lung cancer and 710 population controls of Caucasian or African-American ethnicity in Los Angeles County, California. We did not confirm a strong association between the presence of these inactivating alleles and lung cancer risk [odds ratio (OR) = 0.90, 95% confidence interval (CI) 0.60-1.35 for Caucasians], although there was a small decreased risk among the African-Americans (OR = 0.66, 95% CI 0.38-1.14). Among smokers, when the data are stratified according to lifetime smoking history, there is a suggestion of an association limited to Caucasian smokers of <35 pack-years, the median for all smokers in these data (OR = 0.49, 95% CI 0.23-1.04). However, among African-American smokers, who smoke less than Caucasians, the association did not differ between smoking categories. We also examined the possible role of additional copies of the CYP2D6 gene, which lead to enhanced CYP2D6 activity, in increasing lung cancer risk. Among controls the prevalence of having more than two copies of the CYP2D6 gene and no inactivating alleles was 4.3% for Caucasians and 4.9% for African-Americans. Relative to subjects with an inactivating allele, those with an additional copy of the CYP2D6 gene and no inactivating alleles may be at increased risk of lung cancer, particularly for adenocarcinoma (OR = 3.61, 95% CI 1.08-11.7 for African-Americans and OR = 2.20, 95% CI 0.69-6.0 for Caucasians). Our data suggest that the CYP2D6 genetic polymorphism is not the strong risk factor for lung cancer suggested by some studies of phenotype, but may play a minor role.

Lung cancer risk in relation to the CYP2C9*1/CYP2C9*2 genetic polymorphism among African-Americans and Caucasians in Los Angeles County, California.

CYP2C9 is involved in the metabolism of warfarin and a wide array of other therapeutic agents. It also appears to play a role, along with other cytochrome P450 enzymes, in the metabolism of benzo[a]pyrene, a carcinogen in tobacco smoke. A relatively common allelic variant (termed R144C, Cys144 or more recently CYP2C9*2) has been described that results in the substitution of cysteine for arginine at residue 144 and appears to reduce enzyme activity. We therefore examined the possible association between the presence of the CYP2C9*2 variant allele and risk of lung cancer using peripheral blood DNA from 329 incident cases of lung cancer (152 African-American and 177 Caucasian) and 700 (239 African-American and 461 Caucasian) population controls in Los Angeles County, California. Among the population controls the frequency of the CYP2C9*2 variant allele was lower (p = 0.00002) among African-Americans (0.036) than among Caucasians (0.100). The presence of the CYP2C9*2 variant allele was not associated with a decreased risk of lung cancer; slight but nonstatistically significant elevations in risk were observed for both African-Americans [odds ratio (OR) 1.22, 95% confidence interval (CI) 0.48-3.11] and Caucasians (OR = 1.55, 95% CI 0.96-2.48). The ORs were slightly and nonsignificantly elevated for all histologic types without substantive variation. The association also did not vary materially according to smoking history or whether subjects had the homozygous deletion of the GSTM1 gene. We found no support for the hypothesis that the CYP2C9*2 variant allele decreases the risk of lung cancer. The role of P450s, including CYP2C9, in benzo[a]pyrene metabolism is not fully defined, and CYP2C9 catalyses detoxication as well as activation steps. Thus it is not inconceivable that diminished CYP2C9 activity could increase metabolic activation of benzo[a]pyrene to carcinogenic intermediates. Nonetheless, the small increased risk associated the CYP2C9*2 variant allele in our data is consistent with chance and should not be overinterpreted.

Individual variation in recombination among human males.

Studies of recombination between the markers D6S291 and D6S109 in individuals by sperm typing provide direct evidence for significant variation in recombination among humans. A statistically significant difference in the recombination fraction (range 5.1%-11.2%) was detected among five donors. This variation could reflect polymorphisms in genes affecting recombination or in chromosome structure. Ignoring this variability in studies designed to examine the relationship between physical and genetic distances could lead to incorrect inferences. Individual variation in recombination makes it difficult to predict the recombination fraction for an interval in any particular individual. This could be important in certain genetic counseling situations.

The mutation rate of the human mtDNA deletion mtDNA4977.

The human mitochondrial mutation mtDNA4977 is a 4,977-bp deletion that originates between two 13-bp direct repeats. We grew 220 colonies of cells, each from a single human cell. For each colony, we counted the number of cells and amplified the DNA by PCR to test for the presence of a deletion. To estimate the mutation fate, we used a model that describes the relationship between the mutation rate and the probability that a colony of a given size will contain no mutants, taking into account such factors as possible mitochondrial turnover and mistyping due to PCR error. We estimate that the mutation rate for mtDNA4977 in cultured human cells is 5.95 x 10(-8) per mitochondrial genome replication. This method can be applied to specific chromosomal, as well as mitochondrial, mutations.

Somatic microsatellite mutations as molecular tumor clocks.

Microsatellite (MS) mutations can potentially unravel the past of mutator phenotype tumors, with greater genetic diversity expected in older regions. Rapid clonal expansions of xenografts were characterized by relatively homogenous MS alleles, whereas greater diversity was observed in a colorectal cancer with the greatest variation in its adjacent adenoma. A subcutaneous lung cancer metastasis demonstrated diversity consistent with its one-month clinical duration and evidence of active mitosis during dormancy. The genetic legacy inherent to multistep tumorigenesis provides direct estimates of tumor ages, with up to thousands of cell divisions and high death rates necessary to yield the observed diversities. MS molecular tumor clocks have the unique potential to systematically reconstruct the early and occult evolution of individual human mutator phenotype tumors.

Lung cancer risk in relation to the CYP2E1 Rsa I genetic polymorphism among African-Americans and Caucasians in Los Angeles County.

Genetic polymorphisms in the activation or detoxication of carcinogens, such as those in tobacco smoke, may produce differences in individual susceptibility to lung cancer. The cytochrome P450 CYP2E1 is an enzyme involved in the metabolism of nitrosamines in tobacco smoke. A polymorphism of CYP2E1 detectable by the restriction enzyme Rsa I may be functionally important because it is located in a putative binding site for the transcription factor HNF-1 and has been associated with higher levels of CYP2E1 transcription. It is conceivable that this CYP2E1 Rsa I polymorphism might contribute to differences in susceptibility to lung cancer. We conducted a case-control study of patients with incident lung cancer and population controls in Los Angeles County to examine the association between the CYP2E1 Rsa I polymorphism and lung cancer risk among African-Americans and Caucasians. Samples of white blood cell DNA sufficient for determination of the CYP2E1 Rsa I genotype by a polymerase chain reaction-based assay were obtained from 341 cases and 706 controls with data on lifetime smoking history. No subjects were homozygous for the CYP2E1 Rsa I rare c2 allele. The rare c2 allele was not associated with an increased risk of lung cancer (adjusted odds ratio, OR 0.72; 95% confidence interval, CI = 0.35-1.46). Among the population controls the percentage of subjects carrying the rare c2 allele was lower (p = 0.002) among African-Americans (2%) compared with Caucasians (8%). However, the association between the CYP2E1 Rsa I genotype and lung cancer risk did not differ between ethnic groups. There was no important association between the CYP2E1 Rsa I genotype and lung cancer risk in analyses stratified by cell-type, smoking history, gender, occupational asbestos exposure, and dietary intake of antioxidants vitamin C, vitamin E or beta carotene. Due to the low frequency of the c2 allele in these populations, larger studies would be necessary to rule out a modest association between the CYP2E1 Rsa I polymorphism and lung cancer risk.

Standardization of multiple spirometers at widely separated times and places.

We designed a system for a multiyear longitudinal study of lung function in 12 widely separated communities, intending to minimize variation in instrument-related data. We used multiple rolling-seal spirometer/personal computer systems. Calibrations were checked before, during, and after each day's field testing, using multiple calibration syringes with electronic readouts. The syringes were rotated to obtain data for each syringe-spirometer combination. Before and after each annual field testing season, a laboratory reference spirometer system was calibrated against a water-displacement device and an electronic frequency counter, and then compared against each field spirometer and syringe. Field equipment consistently met American Thoracic Society (ATS) specifications. Variance among spirometers exceeded variance among syringes. A spirometer occasionally changed its volume readout by approximately 1 to 2 %. More rarely, a syringe changed its delivered volume by approximately 1%. Syringes' electronic readouts tracked changes in delivered volume. Syringe readouts were the most stable component of the system, and were more reproducible than the laboratory water-displacement calibration. We conclude that variation in spirometers may limit the reliability of epidemiologic findings, even when these spirometers meet ATS specifications. Frequent calibration checks traceable to an independent standard, and adjustment of individual test results, can reduce measurement error.

Los Angeles study of residential magnetic fields and childhood brain tumors.

A measurement study of residential magnetic fields and brain tumors in children that was added onto an ongoing case-control interview study in Los Angeles County, California, include 298 children under age 20 years with a primary brain tumor diagnosed from 1984 to 1991 and 298 control children identified by random digit dialing. Magnetic fields were determined for all Los Angeles homes where these 596 children lived from conception to diagnosis (1,131 homes) by mapping and coding the wiring configurations outside the home and by taking a series of exterior spot and profile measurements. In addition, for a subset of subjects (35%; 211 homes) 24-hour measurements were taken in the child's room and one other room. Although measured fields are consistently highest in the highest of the five wire code categories, fields in homes in this category are much lower in Los Angeles than in Denver, where the code originated. Brain tumor risk appears not to relate to measured fields inside (p for trend for child's room = 0.98) or outside (p for trend for front wall = 0.82) the home. An apparent increase in risk among children living at diagnosis in homes with underground wiring appears to be an artifact introduced by using current controls for historical cases because this apparent excess risk disappeared in an analysis restricted to the later years of the study when cases and controls were accrued concurrently. Our study does not show an overall association of pediatric brain tumors with measured fields, with "very high" wiring configurations, or with any of several other potential sources of exposure, such as use of various electrical appliances, but the prevalence of high fields (> 2 mG) and very high fields (> 3 mG) in Los Angeles homes was too low to detect a moderate effect of the magnitude reported in other studies.

Lung cancer risk in African-Americans in relation to a race-specific CYP1A1 polymorphism.

The possible association between lung cancer and a polymorphism of the CYP1A1 gene specific to African-Americans was examined using peripheral blood DNA from 144 incident cases of lung cancer and 230 population controls with detailed data on smoking and other risk factors for the disease. The CYP1A1 variant allele was present in 15.2% of controls and 16.7% of cases. The smoking-adjusted odds ratio for the presence of the variant allele in relation to lung cancer risk overall was 1.3 (95% confidence interval, 0.7-2.4). According to histological type, the strongest association was observed for squamous cell carcinoma (odds ratio, 2.1), but this result was compatible with chance (95% confidence interval, 0.8-5.9). Adenocarcinoma was not materially associated with the presence of the variant allele (odds ratio, 1.3; 95% confidence interval, 0.5-3.2). No important associations were observed upon stratification by several risk factors for lung cancer, including smoking history, occupational exposures to asbestos and motor vehicle exhaust, or low intake of the micronutrient antioxidants beta-carotene, vitamin E, or vitamin C. These results do not confirm an earlier report that this CYP1A1 polymorphism may be an important risk factor for adenocarcinoma of the lung in African-Americans.

Polymorphism of glutathione S-transferase M1 and lung cancer risk among African-Americans and Caucasians in Los Angeles County, California.

BACKGROUND: Glutathione S-transferase M1 (GSTM1) is active in the detoxication of a number of carcinogens, including polyaromatic hydrocarbons, such as those present in cigarette smoke. In about 30%-55% of individuals, depending on the ethnic group, there is a virtual absence of GSTM1 enzyme activity due to deletion of both copies of the GSTM1 gene (GSTM1 null genotype). This genetic polymorphism of the GSTM1 gene locus has been proposed as a risk factor for lung cancer. However, results across studies are inconsistent. PURPOSE: We conducted a case-control study of patients with incident lung cancer and population control subjects to examine the association between homozygous deletion of the GSTM1 gene and lung cancer risk among African-Americans and Caucasians. METHODS: At 35 hospitals in Los Angeles County, California, we identified patients with a first diagnosis of lung cancer between September 1, 1990, and January 6, 1994. Of the 859 potentially eligible case patients, 207 had died by the time their physicians had received our request for permission to contact them. We enrolled 356 eligible case patients (167 African-Americans and 189 Caucasians) and 731 eligible control subjects (258 African-Americans and 473 Caucasians, all residents of Los Angeles County). Samples of white blood cell DNA sufficient for determination of the GSTM1 genotype by a polymerase chain reaction-based assay were obtained from 342 case patients and 716 control subjects. The odds ratios (ORs) and 95% confidence intervals (CIs) for lung cancer associated with homozygous deletion of the GSTM1 gene, in total and after stratification by a number of relevant characteristics, were estimated by logistic regression analysis. RESULTS: For patients with all lung cancers combined, the GSTM1 null genotype was associated with an OR of 1.29 (95% CI = 0.94-1.77). The OR was similar among African-Americans (OR = 1.20; 95% CI = 0.72-2.00) and Caucasians (OR = 1.37; 95% CI = 0.91-2.06). The association was strongest for squamous cell carcinoma (OR = 1.57; 95% CI = 0.93-2.63). We observed an OR of 1.77 (95% CI = 1.11-2.82) for the GSTM1 null genotype in relation to lung cancer risk among smokers of less than 40 pack-years, but no association among heavier smokers (OR = 0.90; 95% CI = 0.56-1.44). CONCLUSIONS: Our data do not support a substantial association between homozygous deletion of the GSTM1 gene and the risk of lung cancer overall in this population. However, our data do suggest an elevated risk for lighter smokers with this genotype. IMPLICATIONS: Because the power of our analyses within strata of lifetime smoking history was limited, larger studies will be needed to confirm these findings.

Measurement error in air pollution exposure assessment.

The exposure of an individual to an air pollutant can be assessed indirectly, with a "microenvironmental" approach, or directly with a personal sampler. Both methods of assessment are subject to measurement error, which can cause considerable bias in estimates of health effects. If the exposure estimates are unbiased and the measurement error is nondifferential, the bias in a linear model can be corrected when the variance of the measurement error is known. Unless the measurement error is quite large, estimates of health effects based on individual exposures appear to be more accurate than those based on ambient levels.

Design and analysis of multilevel analytic studies with applications to a study of air pollution.

We discuss a hybrid epidemiologic design that aims to combine two approaches to studying exposure-disease associations. The analytic approach is based on comparisons between individuals, e.g., case-control and cohort studies, and the ecologic approach is based on comparisons between groups. The analytic approach generally provides a stronger basis for inference, in part because of freedom from between-group confounding and better quality data, but the ecologic approach is less susceptible to attenuation bias from measurement error and may provide greater variability in exposure. The design we propose entails selection of a number of groups and enrollment of individuals within each group. Exposures, outcomes, confounders, and modifiers would be assessed on each individual; but additional exposure data might be available on the groups. The analysis would then combine the individual-level and the group-level comparisons, with appropriate adjustments for exposure measurement errors, and would test for compatibility between the two levels of analysis, e.g., to determine whether the associations at the individual level can account for the differences in disease rates between groups. Trade-offs between numbers of groups, numbers of individuals, and the extent of the individual and group measurement protocols are discussed in terms of design efficiency. These issues are illustrated in the context of an on-going study of the health effects of air pollution in southern California, in which 12 communities with different levels and types of pollution have been selected and 3500 school children are being enrolled in a ten-year cohort study.(ABSTRACT TRUNCATED AT 250 WORDS)

Phylogenetic inference: linear invariants and maximum likelihood.

We develop a new statistical method for inferring phylogenies, based on a likelihood ratio test. This method does not require parameter constraints but does require identical evolutionary processes in the sites considered. Another method of phylogenetic inference is the method of linear invariants, described by Cavender (1989, Molecular Biology and Evolution 6, 301-316), based on a notion of Lake (1987, Molecular Biology and Evolution 4, 167-191). We describe a sound mathematical basis for the use of linear invariants. We show that the validity of the method requires parameter constraints, but does not require that the evolutionary processes in differing sites be identical. We show that the method of linear invariants is asymptotically equivalent to a less powerful version of our likelihood ratio test, and is thus essentially a maximum likelihood technique.