Using the entire history in the analysis of nested case cohort samples.

Countermatching designs can provide more efficient estimates than simple matching or case-cohort designs in certain situations such as when good surrogate variables for an exposure of interest are available. We extend pseudolikelihood estimation for the Cox model under countermatching designs to models where time-varying covariates are considered. We also implement pseudolikelihood with calibrated weights to improve efficiency in nested case-control designs in the presence of time-varying variables. A simulation study is carried out, which considers four different scenarios including a binary time-dependent variable, a continuous time-dependent variable, and the case including interactions in each. Simulation results show that pseudolikelihood with calibrated weights under countermatching offers large gains in efficiency if compared to case-cohort. Pseudolikelihood with calibrated weights yielded more efficient estimators than pseudolikelihood estimators. Additionally, estimators were more efficient under countermatching than under case-cohort for the situations considered. The methods are illustrated using the Colorado Plateau uranium miners cohort. Furthermore, we present a general method to generate survival times with time-varying covariates. Copyright © 2016 John Wiley & Sons, Ltd.

Using the whole cohort in the analysis of countermatched samples.

We present a technique for using calibrated weights to incorporate whole-cohort information in the analysis of a countermatched sample. Following Samuelsen's approach for matched case-control sampling, we derive expressions for the marginal sampling probabilities, so that the data can be treated as an unequally-sampled case-cohort design. Pseudolikelihood estimating equations are used to find the estimates. The sampling weights can be calibrated, allowing all whole-cohort variables to be used in estimation; in contrast, the partial likelihood analysis makes use only of a single discrete surrogate for exposure. Using a survey-sampling approach rather than a martingale approach simplifies the theory; in particular, the sampling weights need not be a predictable process. Our simulation results show that pseudolikelihood estimation gives lower efficiency than partial likelihood estimation, but that the gain from calibration of weights can more than compensate for this loss. If there is a good surrogate for exposure, countermatched sampling still outperforms case-cohort and two-phase case-control sampling even when calibrated weights are used. Findings are illustrated with data from the National Wilms' Tumour Study and the Welsh nickel refinery workers study.

Novel loci associated with usual sleep duration: the CHARGE Consortium Genome-Wide Association Study.

Usual sleep duration is a heritable trait correlated with psychiatric morbidity, cardiometabolic disease and mortality, although little is known about the genetic variants influencing this trait. A genome-wide association study (GWAS) of usual sleep duration was conducted using 18 population-based cohorts totaling 47 180 individuals of European ancestry. Genome-wide significant association was identified at two loci. The strongest is located on chromosome 2, in an intergenic region 35- to 80-kb upstream from the thyroid-specific transcription factor PAX8 (lowest P=1.1 × 10(-9)). This finding was replicated in an African-American sample of 4771 individuals (lowest P=9.3 × 10(-4)). The strongest combined association was at rs1823125 (P=1.5 × 10(-10), minor allele frequency 0.26 in the discovery sample, 0.12 in the replication sample), with each copy of the minor allele associated with a sleep duration 3.1 min longer per night. The alleles associated with longer sleep duration were associated in previous GWAS with a more favorable metabolic profile and a lower risk of attention deficit hyperactivity disorder. Understanding the mechanisms underlying these associations may help elucidate biological mechanisms influencing sleep duration and its association with psychiatric, metabolic and cardiovascular disease.

A screening study of drug-drug interactions in cerivastatin users: an adverse effect of clopidogrel.

An analysis of a case-control study of rhabdomyolysis was conducted to screen for previously unrecognized cytochrome P450 enzyme (CYP) 2C8 inhibitors that may cause other clinically important drug-drug interactions. Medication use in cases of rhabdomyolysis using cerivastatin (n = 72) was compared with that in controls using atorvastatin (n = 287) for the period 1998-2001. The use of clopidogrel was strongly associated with rhabdomyolysis (odds ratio (OR) 29.6; 95% confidence interval (CI), 6.1-143). In a replication effort that used the US Food and Drug Administration (FDA) Adverse Event Reporting System (AERS), it was found that clopidogrel was used more commonly in patients with rhabdomyolysis receiving cerivastatin (17%) than in those receiving atorvastatin (0%, OR infinity; 95% CI = 5.2-infinity). Several medications were tested in vitro for their potential to cause drug-drug interactions. Clopidogrel, rosiglitazone, and montelukast were the most potent inhibitors of cerivastatin metabolism. Clopidogrel and its metabolites also inhibited cerivastatin metabolism in human hepatocytes. These epidemiological and in vitro findings suggest that clopidogrel may cause clinically important, dose-dependent drug-drug interactions with other medications metabolized by CYP2C8.

Platelet count and the risk for thrombosis and death in the elderly.

AIM: Our aim was to examine the association between platelet count and the incidence of myocardial infarction, ischemic stroke, hemorrhagic stroke, venous thrombosis, and mortality. METHODS AND RESULTS: Platelet count was measured at baseline in 1989-1990 and at 3 years follow-up, or at baseline (for a newly recruited group) in 1992-1993 in 5766 community-dwelling individuals aged 65 years and older (mean age at baseline, 73 years). During 12-15 years of follow-up, there were 821 incident myocardial infarctions, 807 ischemic strokes, 161 hemorrhagic strokes, 159 venous thrombotic events, and 3413 participants died. Platelet count was not associated with the occurrence of myocardial infarction, ischemic or hemorrhagic stroke, venous thrombosis, or cardiovascular mortality. Non-cardiovascular mortality was higher among both participants with low and with high platelet count. Adjusted non-cardiovascular mortality rates for platelet counts below 100, 100-199, 300-399, and above 400 x 10(9) L(-1) relative to the reference mortality rate in participants with platelet count values between 200 and 299 x 10(9) L(-1) were 1.89 (1.21-2.96), 1.08 (0.98-1.20), 1.20 (1.06-1.37), and 1.47 (1.14-1.90), respectively. CONCLUSION: Platelet counts were not associated with vascular outcomes but low and high platelet counts were associated with non-cardiovascular mortality, including cancer mortality.

ABO genotype and risk of thrombotic events and hemorrhagic stroke.

BACKGROUND: The non-O alleles of the ABO genotype have been associated with an increased risk of thrombosis. Risk associated with the specific A(1), A(2) or B alleles is not well defined. OBJECTIVES: To examine the association of the ABO genotype with myocardial infarction (MI), ischemic stroke, hemorrhagic stroke, and venous thrombosis (VT). PATIENTS AND METHODS: We used data from two ongoing population-based case-control studies of MI, stroke, and VT. Cases included hypertensive adults and postmenopausal women with incident non-fatal MI (n = 1063), ischemic stroke (n = 469), and hemorrhagic stroke (n = 91), and postmenopausal women with incident non-fatal VT (n = 504). Controls were frequency matched to cases on age, sex, hypertension status, and year of identification. ABO genotypes were determined using single-nucleotide polymorphisms, and subjects were grouped by diplotype according to the presence of O(1), O(2), A(11), A(2) and B alleles. Logistic regression was used to test the association of diplotypes with risk of each outcome. RESULTS: As compared with the O(1)O(1) group, the A(11) allele was associated with an increased risk of VT [odds ratio (OR) 1.79; 95% confidence interval (CI) 1.41-2.26] and MI (OR 1.23; 95% CI 1.05-1.44). The B allele was associated with an increased risk of VT (OR 1.82; 95% CI 1.29-2.57) and ischemic stroke (OR 1.59; 95% CI 1.17-2.17). The AB diplotype category was associated with a 2.7-fold risk of VT (OR 2.70; 95% CI 1.73-4.21). No other associations reached significance. CONCLUSIONS: The VT and MI findings are confirmatory, and the ischemic stroke finding with the B allele is a novel finding and needs replication.

Blood pressure control and risk of incident atrial fibrillation.

BACKGROUND: Atrial fibrillation (AF) is a common arrhythmia that affects more than 2 million people in the United States. We sought to determine whether the risk of incident AF among patients treated for hypertension differs by the degree of blood pressure control. METHODS: A population-based, case-control study of 433 patients with verified incident AF and 899 controls was conducted to investigate the relationship between average achieved systolic (SBP) and diastolic (DBP) blood pressure and risk of AF. All patients were members of an integrated health-care delivery system and were pharmacologically treated for hypertension. Medical records were reviewed to confirm the diagnosis of new onset AF and to collect information on medical conditions, health behaviors, and measured blood pressures. Average achieved SBP and DBP were calculated from the three most recent outpatient blood pressure measurements. RESULTS: Compared with the reference level of 120-129 mm Hg, for categories of average achieved SBP of <120, 130-139, 140-149, 150-159, 160-169, and > or =170 mm Hg, the odds ratios (ORs; 95% confidence interval (CI)) for incident AF were 1.99 (1.10, 3.62), 1.19 (0.78, 1.81), 1.40 (0.93, 2.09), 2.02 (1.30, 3.15), 2.27 (1.31, 3.93), and 1.84 (0.89, 3.80), respectively. Based on the population attributable fraction, we estimated that, among patients with treated hypertension, 17.2% (95% CI 4.3%, 28.3%) of incident AF was attributable to an average achieved SBP > or =140 mm Hg. CONCLUSION: Among patients treated for hypertension, uncontrolled elevated SBP and SBP <120 mm Hg were associated with an increased risk of incident AF.

Variation in 24 hemostatic genes and associations with non-fatal myocardial infarction and ischemic stroke.

BACKGROUND: Arterial thrombosis involves platelet aggregation and clot formation, yet little is known about the contribution of genetic variation in fibrin-based hemostatic factors to arterial clotting risk. We hypothesized that common variation in 24 coagulation-fibrinolysis genes would contribute to risk of incident myocardial infarction (MI) or ischemic stroke (IS). METHODS: We conducted a population-based, case-control study. Subjects were hypertensive adults and postmenopausal women 30-79 years of age, who sustained a first MI (n = 856) or IS (n = 368) between 1995 and 2002, and controls matched on age, hypertension status, and calendar year (n = 2,689). We investigated the risk of MI and IS associated with (i) global variation within each gene as measured by common haplotypes and (ii) individual haplotypes and single nucleotide polymorphisms (SNPs). Significance was assessed using a 0.2 threshold of the false discovery rate q-value, which accounts for multiple testing. RESULTS: After accounting for multiple testing, global genetic variation in factor (F) VIII was associated with IS risk. Two haplotypes in FVIII and one in FXIIIa1 were significantly associated with increased IS risk (all q-values < 0.2). A plasminogen gene SNP was associated with MI risk. All are new discoveries not previously reported. Another 24 tests had P-values < 0.05 and q-values > 0.2 in MI and IS analyses, 23 of which are new and hypothesis generating. CONCLUSIONS: Apart from the association of FVIII variation with IS, we found little evidence that common variation in the 24 candidate fibrin-based hemostasis genes strongly influences arterial thrombotic risk, but our results cannot rule out small effects.

The differential association of conjugated equine estrogen and esterified estrogen with activated protein C resistance in postmenopausal women.

OBJECTIVES: Clinical trials have demonstrated that oral conjugated equine estrogen (CEE) therapy with or without medroxyprogesterone (MPA) increases venous thrombotic risk but this safety issue has not been investigated for other oral estrogens. Based on observational study findings that esterified estrogen (EE) was not associated with venous thrombotic risk whereas CEE was, we hypothesized that CEE users would be more resistant to activated protein C (APC), a prothrombotic phenotype, than EE users. METHODS: We conducted an observational, cross-sectional study of postmenopausal women 30-89 years old who were controls in a case-control study of venous thrombosis. Use of CEE, EE, and MPA at the time of phlebotomy was determined using computerized pharmacy records. APC resistance was measured in plasma by the endogenous thrombin potential normalized APC sensitivity ratio. Adjusted mean APC resistance values were compared across estrogen type and CEE:EE ratios are presented. RESULTS: There were 119 CEE and 92 EE users at the time of phlebotomy. Compared with EE users, CEE users had APC resistance measures that were 52% higher (1.52; 95% confidence intervals: 1.07-2.17) in adjusted analyses. Restricting to modal dose users (0.625 mg) and stratifying by MPA use did not materially change associations. CONCLUSIONS: CEE use was associated with higher levels of APC resistance when compared with EE use in postmenopausal women. These findings might provide an explanation for the higher risk of venous thromboembolism previously observed with CEE compared with EE use and, if replicated, may have safety implications for women when choosing an estrogen for symptom relief.

Respiratory muscle strength and the risk of incident cardiovascular events.

BACKGROUND: Maximal inspiratory pressure (MIP) is a measure of inspiratory muscle strength. The prognostic importance of MIP for cardiovascular events among elderly community dwelling individuals is unknown. Diminished forced vital capacity (FVC) is a risk factor for cardiovascular events which remains largely unexplained. METHODS: MIP was measured at the baseline examination of the Cardiovascular Health Study. Participants had to be free of prevalent congestive heart failure (CHF), myocardial infarction (MI), and stroke. RESULTS: Subjects in the lowest quintile of MIP had a 1.5-fold increased risk of MI (HR 1.48, 95% CI 1.07 to 2.06) and cardiovascular disease (CVD) death (HR 1.54, 95% CI 1.09 to 2.15) after adjustment for non-pulmonary function covariates. There was a potential inverse relationship with stroke (HR 1.36, 95% CI 0.97 to 1.90), but there was little evidence of an association between MIP and CHF (HR 1.22, 95% CI 0.93 to 1.60). The addition of FVC to models attenuated the HR associated with MIP only modestly; similarly, addition of MIP attenuated the HR associated with FVC only modestly. CONCLUSIONS: A reduced MIP is an independent risk factor for MI and CVD death, and a suggestion of an increased risk for stroke. This association with MIP appeared to be mediated through mechanisms other than inflammation.

Measurement of offline exhaled nitric oxide in a study of community exposure to air pollution.

As part of a large panel study in Seattle, Washington, we measured levels of exhaled nitric oxide (eNO) in children's homes and fixed-site particulate matter with aerodynamic diameters of 2.5 micro m or less (PM(2.5)) outside and inside the homes as well as personal PM(2.5) during winter and spring sessions of 2000-2001. Nineteen subjects 6-13 years of age participated; 9 of the 19 were on inhaled corticosteroid (ICS) therapy. Exhaled breath measurements were collected offline into a Mylar balloon for up to 10 consecutive days. Mean eNO values were 19.1 (SD +/- 11.4) ppb in winter sessions and 12.5 +/- 6.6 ppb in spring sessions. Fixed-site PM(2.5) mean concentrations were 10.1 +/- 5.7 microg/m(3) outside homes and 13.3 +/- 1.4 inside homes; the personal PM(2.5) mean was 13.4 +/- 3.2 microg/m(3). We used a linear mixed-effects model with random intercept and an interaction term for medications to test for within-subject-within-session associations between eNO and various PM(2.5) values. We found a 10 microg/m(3) increase in PM(2.5) from the outdoor, indoor, personal, and central-site measurements that was associated with increases in eNO in all subjects at lag day zero. The effect was 4.3 ppb [95% confidence interval (CI), 1.4-7.29] with the outdoor monitor, 4.2 ppb (95% CI, 1.02-7.4) for the indoor monitor, 4.5 ppb (95% CI, 1.02-7.9) with the personal monitor, and 3.8 ppb (95% CI, 1.2-6.4) for the central monitors. The interaction term for medication category (ICS users vs. nonusers) was significant in all analyses. These findings suggest that eNO can be used as an assessment tool in epidemiologic studies of health effects of air pollution.

Combining components of quality of life to increase precision and evaluate trade-offs.

Methods for combining measurements on multiple dimensions of quality of life can reduce the dimensionality of the data and increase the precision of estimation. When the dimensions are weighted according to their importance to patients, the resulting estimate is clinically useful and provides a step towards a true utility estimate. We derive two such weighting methods using linear regression on a measure of overall quality of life and demonstrate their usefulness in the analysis of quality of life data from two clinical trials of cancer therapies. Procedures for transforming the quality of life measures into utility measures are demonstrated.

Association between blood pressure level and the risk of myocardial infarction, stroke, and total mortality: the cardiovascular health study.

BACKGROUND: Recent reports have drawn attention to the importance of pulse pressure as a predictor of cardiovascular events. Pulse pressure is used neither by clinicians nor by guidelines to define treatable levels of blood pressure. METHODS: In the Cardiovascular Health Study, 5888 adults 65 years and older were recruited from 4 US centers. At baseline in 1989-1990, participants underwent an extensive examination, and all subsequent cardiovascular events were ascertained and classified. RESULTS: At baseline, 1961 men and 2941 women were at risk for an incident myocardial infarction or stroke. During follow-up that averaged 6.7 years, 572 subjects had a coronary event, 385 had a stroke, and 896 died. After adjustment for potential confounders, systolic blood pressure (SBP), diastolic blood pressure (DBP), and pulse pressure were directly associated with the risk of incident myocardial infarction and stroke. Only SBP was associated with total mortality. Importantly, SBP was a better predictor of cardiovascular events than DBP or pulse pressure. In the adjusted model for myocardial infarction, a 1-SD change in SBP, DBP, and pulse pressure was associated with hazard ratios (95% confidence intervals) of 1.24 (1.15-1.35), 1.13 (1.04-1.22), and 1.21 (1.12-1.31), respectively; and adding pulse pressure or DBP to the model did not improve the fit. For stroke, the hazard ratios (95% confidence intervals) were 1.34 (1.21-1.47) with SBP, 1.29 (1.17-1.42) with DBP, and 1.21 (1.10-1.34) with pulse pressure. The association between blood pressure level and cardiovascular disease risk was generally linear; specifically, there was no evidence of a J-shaped relationship. In those with treated hypertension, the hazard ratios for the association of SBP with the risks for myocardial infarction and stroke were less pronounced than in those without treated hypertension. CONCLUSION: In this population-based study of older adults, although all measures of blood pressure were strongly and directly related to the risk of coronary and cerebrovascular events, SBP was the best single predictor of cardiovascular events.

Referent selection in case-crossover analyses of acute health effects of air pollution.

The case-crossover design was proposed for the study of a transient effect of an intermittent exposure on the subsequent occurrence of a rare acute-onset disease. This design can be an alternative to Poisson time series regression for studying the health effects of fine particulate matter air pollution. Characteristics of time-series of particulate matter, including long-term time trends, seasonal trends, and short-term autocorrelations, require that referent selection in the case-crossover design be considered carefully and adapted to minimize bias. We performed simulations to evaluate the bias associated with various referent selection strategies for a proposed case-crossover study of associations between particulate matter and primary cardiac arrest. Some a priori reasonable strategies were associated with a relative bias as large as 10%, but for most strategies the relative bias was less than 2% with confidence interval coverage within 1% of the nominal level. We show that referent selection for case-crossover designs raises the same issues as selection of smoothing method for time series analyses. In addition, conditional logistic regression analysis is not strictly valid for some case-crossover designs, introducing further bias.

A case-crossover analysis of particulate matter air pollution and out-of-hospital primary cardiac arrest.

Numerous epidemiologic studies have reported increases in the daily incidence of cardiovascular mortality and morbidity associated with increases in daily levels of particulate matter air pollution. We studied the association between the incidence of primary cardiac arrest and two daily measures of particulate matter using a case-crossover study of 362 cases of out-of-hospital cardiac arrest. All cases were attended by paramedics and had no history of clinically recognized heart disease or life-threatening comorbidities. We compared particulate matter levels at index times with particulate matter levels from referent days matched on day of week within strata defined by month and year. The estimated relative risk at a lag of 1 day for an interquartile range (IQR) change in nephelometry (0.51 x 10(-1) km(-1)) was 0.893 (95% CI = 0.779-1.024). The estimated relative risk at a lag of 1 day for an IQR change in PM10 (19.3 microg m(-3)) was 0.868 (95% CI = 0.744-1.012). Other lag periods gave similar results. We did not find evidence of confounding by carbon monoxide or sulfur dioxide. Analysis of effect modification by individual-level variables did not reveal any susceptible subgroups. These findings do no support an association between particulate matter and increased risk of primary cardiac arrest among persons without clinically recognized heart disease. The null results of this study may result from several factors, including the highly selected nature of this case series and the relatively low particulate matter levels in the Seattle metropolitan area.

Time-dependent ROC curves for censored survival data and a diagnostic marker.

ROC curves are a popular method for displaying sensitivity and specificity of a continuous diagnostic marker, X, for a binary disease variable, D. However, many disease outcomes are time dependent, D(t), and ROC curves that vary as a function of time may be more appropriate. A common example of a time-dependent variable is vital status, where D(t) = 1 if a patient has died prior to time t and zero otherwise. We propose summarizing the discrimination potential of a marker X, measured at baseline (t = 0), by calculating ROC curves for cumulative disease or death incidence by time t, which we denote as ROC(t). A typical complexity with survival data is that observations may be censored. Two ROC curve estimators are proposed that can accommodate censored data. A simple estimator is based on using the Kaplan-Meier estimator for each possible subset X > c. However, this estimator does not guarantee the necessary condition that sensitivity and specificity are monotone in X. An alternative estimator that does guarantee monotonicity is based on a nearest neighbor estimator for the bivariate distribution function of (X, T), where T represents survival time (Akritas, M. J., 1994, Annals of Statistics 22, 1299-1327). We present an example where ROC(t) is used to compare a standard and a modified flow cytometry measurement for predicting survival after detection of breast cancer and an example where the ROC(t) curve displays the impact of modifying eligibility criteria for sample size and power in HIV prevention trials.

A comparison of standard gamble, time trade-off, and adjusted time trade-off scores.

OBJECTIVE: To adjust patients' time trade-off (TTO) scores using information on their utility functions for survival time to derive a measure of health state utility equivalent to the standard gamble (SG). METHODS: A sample of 199 cardiovascular patients were asked three TTO and SG questions (to assess their own health state), and three certainty equivalent questions (to assess their utility function for survival time) in an interview. RESULTS: Patient's utility functions for time were increasingly concave, but being unable to model this successfully, a constant function with an averaged level of concavity was used. The raw TTO scores were significantly higher than SG scores, while the adjusted TTO scores were equivalent to the SG. CONCLUSIONS: Raw time trade-off scores will give biased estimates of health state utility when patients' utility functions for time are not linear, but these can be adjusted to yield true utilities. The constant proportional risk-posture assumption of the conventional QALY model, on which previous attempts to adjust time trade-offs have been based, was not supported by the data.

Variation in the quality of lumbar spine MR images in Washington State.

PURPOSE: To investigate the variation in quality of lumbar spine magnetic resonance (MR) images as a function of type of ownership of the imaging center, number of studies performed per month, specialty training of the image interpreter, and field strength of the MR unit. MATERIALS AND METHODS: Data were collected from all imaging facilities in western Washington state that received reimbursement from the Washington State Health Care Authority. Three readers with expertise in spine imaging, who were blinded to center and patient identification information, rated the technical image quality of each study. All MR images of the lumbar spine (maximum of six) paid for by the health care authority were evaluated. If a center had performed more than six studies, then six were randomly selected for evaluation. RESULTS: Variation in quality scores among sites was significant (P =.001). Field strength was the strongest predictor of better quality. Poorer quality was associated with for-profit ownership, a larger number of radiologists at the site reading MR images, and a larger percentage of studies checked by a radiologist prior to the end of the examination. CONCLUSION: There was significant variation in the quality of MR images of the lumbar spine, and at least a portion of this variation was attributable to characteristics of the imaging center.

Effects of ambient air pollution on symptoms of asthma in Seattle-area children enrolled in the CAMP study.

We observed a panel of 133 children (5-13 years of age) with asthma residing in the greater Seattle, Washington, area for an average of 58 days (range 28-112 days) during screening for enrollment in the Childhood Asthma Management Program (CAMP) study. Daily self-reports of asthma symptoms were obtained from study diaries and compared with ambient air pollution levels in marginal repeated measures logistic regression models. We defined days with asthma symptoms as any day a child reported at least one mild asthma episode. All analyses were controlled for subject-specific variables [age, race, sex, baseline height, and FEV(1) PC(20) concentration (methacholine provocative concentration required to produce a 20% decrease in forced expiratory volume in 1 sec)] and potential time-dependent confounders (day of week, season, and temperature). Because of variable observation periods for participants, we estimated both between- and within-subject air pollutant effects. Our primary interest was in the within-subject effects: the effect of air pollutant excursions from typical levels in each child's observation period on the odds of asthma symptoms. In single-pollutant models, the population average estimates indicated a 30% [95% confidence interval (CI), 11-52%] increase for a 1-ppm increment in carbon monoxide lagged 1 day, an 18% (95% CI, 5-33%) increase for a 10-microg/m(3) increment in same-day particulate matter < 1.0 microm (PM(1.0)), and an 11% (95% CI, 3-20%) increase for a 10-microg/m(3) increment in particulate matter < 10 microm (PM(10)) lagged 1 day. Conditional on the previous day's asthma symptoms, we estimated 25% (95% CI, 10-42%), 14% (95% CI, 4-26%), and 10% (95% CI, 3-16%) increases in the odds of asthma symptoms associated with increases in CO, PM(1.0), and PM(10), respectively. We did not find any association between sulfur dioxide (SO(2)) and the odds of asthma symptoms. In multipollutant models, the separate pollutant effects were smaller. The overall effect of an increase in both CO and PM(1. 0) was a 31% (95% CI, 11-55%) increase in the odds of symptoms of asthma. We conclude that there is an association between change in short-term air pollution levels, as indexed by PM and CO, and the occurrence of asthma symptoms among children in Seattle. Although PM effects on asthma have been found in other studies, it is likely that CO is a marker for vehicle exhaust and other combustion by-products that aggravate asthma.