Showing papers on "Confidence interval published in 2000"

Vitamin E supplementation and cardiovascular events in high-risk patients.

Abstract: Observational and experimental studies suggest that the amount of vitamin E ingested in food and in supplements is associated with a lower risk of coronary heart disease and atherosclerosis.We enrolled a total of 2545 women and 6996 men 55 years of age or older who were at high risk for cardiovascular events because they had cardiovascular disease or diabetes in addition to one other risk factor. These patients were randomly assigned according to a two-by-two factorial design to receive either 400 IU of vitamin E daily from natural sources or matching placebo and either an angiotensin-converting-enzyme inhibitor (ramipril) or matching placebo for a mean of 4.5 years (the results of the comparison of ramipril and placebo are reported in a companion article). The primary outcome was a composite of myocardial infarction, stroke, and death from cardiovascular causes. The secondary outcomes included unstable angina, congestive heart failure, revascularization or amputation, death from any cause, complications of diabetes, and cancer.A total of 772 of the 4761 patients assigned to vitamin E (16.2 percent) and 739 of the 4780 assigned to placebo (15.5 percent) had a primary outcome event (relative risk, 1.05; 95 percent confidence interval, 0.95 to 1.16; P=0.33). There were no significant differences in the numbers of deaths from cardiovascular causes (342 of those assigned to vitamin E vs. 328 of those assigned to placebo; relative risk, 1.05; 95 percent confidence interval, 0.90 to 1.22), myocardial infarction (532 vs. 524; relative risk, 1.02; 95 percent confidence interval, 0.90 to 1.15), or stroke (209 vs. 180; relative risk, 1.17; 95 percent confidence interval, 0.95 to 1.42). There were also no significant differences in the incidence of secondary cardiovascular outcomes or in death from any cause. There were no significant adverse effects of vitamin E.In patients at high risk for cardiovascular events, treatment with vitamin E for a mean of 4.5 years had no apparent effect on cardiovascular outcomes.

Marginal structural models to estimate the causal effect of zidovudine on the survival of HIV-positive men.

Abstract: Standard methods for survival analysis, such as the time-dependent Cox model, may produce biased effect estimates when there exist time-dependent confounders that are themselves affected by previous treatment or exposure. Marginal structural models are a new class of causal models the parameters of which are estimated through inverse-probability-of-treatment weighting; these models allow for appropriate adjustment for confounding. We describe the marginal structural Cox proportional hazards model and use it to estimate the causal effect of zidovudine on the survival of human immunodeficiency virus-positive men participating in the Multicenter AIDS Cohort Study. In this study, CD4 lymphocyte count is both a time-dependent confounder of the causal effect of zidovudine on survival and is affected by past zidovudine treatment. The crude mortality rate ratio (95% confidence interval) for zidovudine was 3.6 (3.0-4.3), which reflects the presence of confounding. After controlling for baseline CD4 count and other baseline covariates using standard methods, the mortality rate ratio decreased to 2.3 (1.9-2.8). Using a marginal structural Cox model to control further for time-dependent confounding due to CD4 count and other time-dependent covariates, the mortality rate ratio was 0.7 (95% conservative confidence interval = 0.6-1.0). We compare marginal structural models with previously proposed causal methods.

Statistics with confidence

Abstract: Uncertainty and Psychological Research Variables and Measurement Exploring, Describing, Displaying and Summarizing Research Design and Probability Sampling Distributions and Confidence Intervals Statistical Models and Significance Tests Predicting a Quantitative Variable from a Categorical Variable The t Test and Analysis of Variance Quantitative Predictors Regression and Correlation Predicting Categorical Variables Contingency Tables and Chi-Square More Than Two Variables A Peek at Multivariate Analysis Putting Statistics into Perspective

Bootstrap confidence intervals : when, which, what? A practical guide for medical statisticians

Abstract: Since the early 1980s, a bewildering array of methods for constructing bootstrap confidence intervals have been proposed. In this article, we address the following questions. First, when should bootstrap confidence intervals be used. Secondly, which method should be chosen, and thirdly, how should it be implemented. In order to do this, we review the common algorithms for resampling and methods for constructing bootstrap confidence intervals, together with some less well known ones, highlighting their strengths and weaknesses. We then present a simulation study, a flow chart for choosing an appropriate method and a survival analysis example.

Statistics with Confidence : Confidence Intervals and Statistical Guidelines

Abstract: Statistics with Confidence is a widely acclaimed classic introduction to confidence intervals. The second edition, with contributions from leading medical statisticians, has been thoroughly revised and expanded. New chapters include: * diagnostic studies. * clinical trials and meta analyses. * confidence intervals and sample sizes. The Windows computer software (CIA) in the accompanying disk enables confidence intervals to be calculated using the methods described in the book. This comprehensive collection of methods includes some that are not readily available elsewhere. Featuring methods for using confidence intervals, illustrative worked examples and helpful checklists this is a truly practical guide for clinical readers to a fundamental aspect of medical statistics.

Statistics Notes: The odds ratio

Abstract: In recent years odds ratios have become widely used in medical reports—almost certainly some will appear in today's BMJ . There are three reasons for this. Firstly, they provide an estimate (with confidence interval) for the relationship between two binary (“yes or no”) variables. Secondly, they enable us to examine the effects of other variables on that relationship, using logistic regression. Thirdly, they have a special and very convenient interpretation in case-control studies (dealt with in a future note). The odds are a way of representing probability, especially familiar for betting. For example, the odds that a single throw of a die will produce a six are 1 to 5, or 1/5. The odds is the ratio of the probability that the event of interest occurs to the probability that it does not. This is often estimated by the ratio of the number of times that the event of interest occurs to …

All-cause mortality associated with physical activity during leisure time, work, sports, and cycling to work

Abstract: BACKGROUND: Physical activity is associated with low mortality in men, but little is known about the association in women, different age groups, and everyday activity. OBJECTIVE: To evaluate the relationship between levels of physical activity during work, leisure time, cycling to work, and sports participation and all-cause mortality. DESIGN: Prospective study to assess different types of physical activity associated with risk of mortality during follow-up after the subsequent examination. Mean follow-up from examination was 14.5 years. SETTING: Copenhagen University Hospital, Copenhagen, Denmark. PARTICIPANTS: Participants were 13,375 women and 17,265 men, 20 to 93 years of age, who were randomly selected. Physical activity was assessed by self-report, and health status, including blood pressure, total cholesterol level, triglyceride levels, body mass index, smoking, and educational level, was evaluated. MAIN OUTCOME MEASURE: All-cause mortality. RESULTS: A total of 2,881 women and 5,668 men died. Compared with the sedentary, age- and sex-adjusted mortality rates in leisure time physical activity groups 2 to 4 were 0.68 (95% confidence interval, 0.64-0.71), 0.61 (95% confidence interval, 0.57-0.66), and 0.53 (95% confidence interval, 0.41-0.68), respectively, with no difference between sexes and age groups. Within the moderately and highly active persons, sports participants experienced only half the mortality of nonparticipants. Bicycling to work decreased risk of mortality in approximately 40% after multivariate adjustment, including leisure time physical activity. CONCLUSIONS: Leisure time physical activity was inversely associated with all-cause mortality in both men and women in all age groups. Benefit was found from moderate leisure time physical activity, with further benefit from sports activity and bicycling as transportation.

Breast Density as a Predictor of Mammographic Detection: Comparison of Interval- and Screen-Detected Cancers

Abstract: Background Screening mammography is the best method to reduce mortality from breast cancer, yet some breast cancers cannot be detected by mammography. Cancers diagnosed after a negative mammogram are known as interval cancers. This study investigated whether mammographic breast density is related to the risk of interval cancer. Methods Subjects were selected from women participating in mammographic screening from 1988 through 1993 in a large health maintenance organization based in Seattle, WA. Women were eligible for the study if they had been diagnosed with a first primary invasive breast cancer within 24 months of a screening mammogram and before a subsequent one. Interval cancer case subjects (n = 149) were women whose breast cancer occurred after a negative or benign mammographic assessment. Screen-detected control subjects (n = 388) were diagnosed after a positive screening mammogram. One radiologist, who was blinded to cancer status, assessed breast density by use of the American College of Radiology Breast Imaging Reporting and Data System. Results Mammographic sensitivity (i.e., the ability of mammography to detect a cancer) was 80% among women with predominantly fatty breasts but just 30% in women with extremely dense breasts. The odds ratio (OR) for interval cancer among women with extremely dense breasts was 6.14 (95% confidence interval [CI] = 1.95-19.4), compared with women with extremely fatty breasts, after adjustment for age at index mammogram, menopausal status, use of hormone replacement therapy, and body mass index. When only those interval cancer cases confirmed by retrospective review of index mammograms were considered, the OR increased to 9.47 (95% CI = 2.78-32.3). Conclusion Mammographic breast density appears to be a major risk factor for interval cancer.

Associations of general and abdominal obesity with multiple health outcomes in older women: the Iowa Women's Health Study.

Abstract: confidence interval, 11-14), compared with 091 (95% confidence interval, 08-10) for BMI and 11 (95% confidence interval, 10-13) for waist circumference The waist-hip ratio was also associated positively with mortality from coronary heart disease, other cardiovascular diseases, cancer, and other causes The waist-hip ratio was associated less consistently than BMI or waist circumference with cancer incidence All anthropometric indexes were associated with incidence of diabetes and hypertension For example, women simultaneously in the highest quintiles of BMI and waist-hip ratio had a relative risk of diabetes of 29 (95% confidence interval, 18-46) vs women in the lowest combined quintiles Conclusion: The waist-hip ratio offers additional prognostic information beyond BMI and waist circumference

Sample size requirements for estimating pearson, kendall and spearman correlations

Abstract: Interval estimates of the Pearson, Kendall tau-a and Spearman correlations are reviewed and an improved standard error for the Spearman correlation is proposed. The sample size required to yield a confidence interval having the desired width is examined. A two-stage approximation to the sample size requirement is shown to give accurate results.

Fractal Correlation Properties of R-R Interval Dynamics and Mortality in Patients With Depressed Left Ventricular Function After an Acute Myocardial Infarction

Abstract: BACKGROUND: Preliminary data suggest that the analysis of R-R interval variability by fractal analysis methods may provide clinically useful information on patients with heart failure. The purpose of this study was to compare the prognostic power of new fractal and traditional measures of R-R interval variability as predictors of death after acute myocardial infarction. METHODS AND RESULTS: Time and frequency domain heart rate (HR) variability measures, along with short- and long-term correlation (fractal) properties of R-R intervals (exponents alpha(1) and alpha(2)) and power-law scaling of the power spectra (exponent beta), were assessed from 24-hour Holter recordings in 446 survivors of acute myocardial infarction with a depressed left ventricular function (ejection fraction

Effect of Hormone Replacement Therapy on Breast Cancer Risk: Estrogen Versus Estrogen Plus Progestin

Abstract: Background: Hormone replacement therapy (HRT) given as unopposed estrogen replacement therapy (ERT) gained widespread popularity in the United States in the 1960s and 1970s. Recent prescribing practices have favored combination HRT (CHRT), i.e., adding a progestin to estrogen for the entire monthly cycle (continuous combined replacement therapy [CCRT]) or a part of the cycle (sequential estrogen plus progestin therapy [SEPRT]). Few data exist on the association between CHRT and breast cancer risk. We determined the effects of CHRT on a woman’s risk of developing breast cancer in a population-based, case‐control study. Methods: Case subjects included those with incident breast cancers diagnosed over 4 1 ⁄ 2 years in Los Angeles County, CA, in the late 1980s and 1990s. Control subjects were neighborhood residents who were individually matched to case subjects on age and race. Case subjects and control subjects were interviewed in person to collect information on known breast cancer risk factors as well as on HRT use. Information on 1897 postmenopausal case subjects and on 1637 postmenopausal control subjects aged 55‐72 years who had not undergone a simple hysterectomy was analyzed. Breast cancer risks associated with the various types of HRT were estimated as odds ratios (ORs) after adjusting simultaneously for the different forms of HRT and for known risk factors of breast cancer. All P values are two-sided. Results: HRT was associated with a 10% higher breast cancer risk for each 5 years of use (OR 5 = 1.10; 95% confidence interval [CI] = 1.02‐1.18). Risk was substantially higher for CHRT use (OR5 = 1.24; 95% CI = 1.07‐1.45) than for ERT use (OR 5 = 1.06; 95% CI = 0.97‐1.15). Risk estimates were higher for SEPRT (OR 5 = 1.38; 95% CI = 1.13‐1.68) than for CCRT (OR5 = 1.09; 95% CI = 0.88‐1.35), but this difference was not statistically significant. Conclusions: This study provides strong evidence that the addition of a progestin to HRT enhances markedly the risk of breast cancer relative to estrogen use alone. These findings have important implications for the risk‐ benefit equation for HRT in women using CHRT. [J Natl Cancer Inst 2000; 92:328‐32]

Survival, Functional Limitations, and Self-rated Health in the NHANES I Epidemiologic Follow-up Study, 1992

Abstract: This study examined relative hazards for mortality and functional limitations according to poor self-ratings of health using prospective data from the NHANES I Epidemiologic Follow-up Study, a representative sample of US adults aged 25-74 years that has been followed since the First National Health and Nutrition Examination Survey (NHANES I) was conducted in 1971-1975. Follow-up data were taken from death records and from the 1982 and 1992 reinterviews. Respondents (n = 6,913) provided extensive baseline data through physician examinations, laboratory testing, and self-reports of conditions, symptoms, and risk behaviors. Functional limitations were assessed among survivors in 1982 and 1992. Cox regression models accounting for sample design indicated that baseline self-rated health was associated with a significantly reduced hazard of mortality for males but not for females through 1992; adjusted hazards ratios for excellent health as compared with poor health were 0.52 for males (95% confidence interval: 0.36, 0.73) and 0.80 for females (95% confidence interval: 0.51, 1.23). Self-rated health also predicted 1982 and 1992 functional limitation for both men and women and 1992 function net of 1982 function for men only. Self-rated health contributes unique information to epidemiologic studies that is not captured by standard clinical assessments or self-reported histories, but evidence suggests that the effect may be stronger for men than for women.

Effect of long-term treatment with inhaled budesonide on adult height in children with asthma.

Abstract: Background Short-term studies have shown that inhaled corticosteroids may reduce the growth of children with asthma. However, the effect of long-term treatment on adult height is uncertain. Methods We conducted a prospective study in children with asthma to examine the effect of long-term treatment with inhaled budesonide on adult height. We report on 211 children who have attained adult height: 142 budesonide-treated children with asthma, 18 control patients with asthma who have never received inhaled corticosteroids, and 51 healthy siblings of patients in the budesonide group, who also served as controls. Results The children in the budesonide group attained adult height after a mean of 9.2 years of budesonide treatment (range, 3 to 13) at a mean daily dose of 412 μg (range, 110 to 877). The mean cumulative dose of budesonide was 1.35 g (range, 0.41 to 3.99). The mean differences between the measured and target adult heights were +0.3 cm (95 percent confidence interval, –0.6 to +1.2) for the budesonide-...

Consumption of NSAIDs and the development of congestive heart failure in elderly patients: an underrecognized public health problem.

Abstract: Methods: We conducted a matched case-control study of the relationship between recent use of NSAIDs and hospitalization with CHF. Cases (n = 365) were patients admitted to hospitals with a primary diagnosis of CHF. Controls (n = 658) were patients without CHF who were admitted to the same hospitals as case patients. Structured interviews were used to obtain information on several study factors, including recent use of aspirin and other NSAIDs. Results: Use of NSAIDs (other than low-dose aspirin) in the previous week was associated with a doubling of the odds of a hospital admission with CHF (adjusted odds ratio, 2.1; 95% confidence interval, 1.2-3.3). Use of NSAIDs by patients with a history of heart disease was associated with an odds ratio of 10.5 (95% confidence interval, 2.5-44.9) for first admission with heart failure, compared with 1.6 (95% confidence interval, 0.7-3.7) in those without such a history. The odds of a first admission to a hospital with CHF was positively related to the dose of NSAID consumed in the previous week, and was increased to a greater extent with long half-life than with short half-life drugs. Assuming these relationships are causal, NSAIDs were responsible for approximately 19% of hospital admissions with CHF. Conclusions: The burden of illness resulting from NSAID-related CHF may exceed that resulting from gastrointestinal tract damage. NSAIDs should be used with caution in patients with a history of cardiovascular disease. Arch Intern Med. 2000;160:777-784

Behavioral Treatment for Chronic Low Back Pain : A Systematic Review Within the Framework of the Cochrane Back Review Group

Abstract: Study Design. A systematic review of randomized controlled trials. Summary of Background Data. The treatment of chronic low back pain is not primarily focused on removing an underlying organic disease but at the reduction of disability through the modification of environmental contingencies and cognitive processes. Behavioral interventions are commonly used in the treatment of chronic (disabling) low back pain. Objectives. To determine whether behavioral therapy is more effective than reference treatments for chronic nonspecific low back pain and which type of behavioral treatment is most effective. Methods. The authors searched the Medline and PsychLit databases and the Cochrane Controlled Trials Register up to April 1999. and Embase up to September 1999. Also screened were references of identified randomized trials and relevant systematic reviews. Methodologic quality assessment and data extraction were performed independently by two reviewers. The magnitude of effect was assessed by computing a pooled effect size for each domain (i.e., behavioral outcomes, overall improvement, back pain-specific and generic functional status, return to work, and pain intensity) using the random effects model. Results. Only six (25%) studies were high quality. There is strong evidence (level 1) that behavioral treatment has a moderate positive effect on pain intensity (pooled effect size 0.62; 95% confidence interval [CI] 0.25, 0.98), and small positive effects on generic functional status (pooled effect size 0.35; 95% CI: 0.04, 0.74) and behavioral outcomes (pooled effect size 0.40; 95% CI: 0.10, 0.70) of patients with chronic low back pain when compared-with waiting-list controls or no treatment. There is moderate evidence (level 2) that a addition of behavioral component to a usual treatment program for chronic low backpain has no positive short-term effect on generic functional status (pooled effect size 0.31; 95% Cl: 0.01, 0.64), pain intensity (pooled effect size 0.03; 95% CI: 0.30, 0.36), and behavioral outcomes (pooled effect size 0.19; 95% CI: 0.08, 0.45). Conclusions. Behavioral treatment seems to be an effective treatment for patients with chronic low back pain,but it is still unknown what type of patients benefitmost from what type of behavioral treatment.

Maternal morbidity and mortality associated with interpregnancy interval: cross sectional study

Abstract: Objective: To study the impact of interpregnancy interval on maternal morbidity and mortality. Design: Retrospective cross sectional study with data from the Perinatal Information System database of the Latin American Centre for Perinatology and Human Development, Montevideo, Uruguay. Setting: Latin America and the Caribbean, 1985-97. Participants: 456 889 parous women delivering singleton infants. Main outcome measures: Crude and adjusted odds ratios of the effects of short and long interpregnancy intervals on maternal death, pre-eclampsia, eclampsia, gestational diabetes mellitus, third trimester bleeding, premature rupture of membranes, postpartum haemorrhage, puerperal endometritis, and anaemia. Results: Short ( 59 months) interpregnancy intervals were observed for 2.8% and 19.5% of women, respectively. After adjustment for major confounding factors, compared with those conceiving at 18 to 23 months after a previous birth, women with interpregnancy intervals of 5 months or less had higher risks for maternal death (odds ratio 2.54; 95% confidence interval 1.22 to 5.38), third trimester bleeding (1.73; 1.42 to 2.24), premature rupture of membranes (1.72; 1.53 to 1.93), puerperal endometritis (1.33; 1.22 to 1.45), and anaemia (1.30; 1.18 to 1.43). Compared with women with interpregnancy intervals of 18 to 23 months, women with interpregnancy intervals longer than 59 months had significantly increased risks of pre-eclampsia (1.83; 1.72 to 1.94) and eclampsia (1.80; 1.38 to 2.32). Conclusions: Interpregnancy intervals less than 6 months and longer than 59 months are associated with an increased risk of adverse maternal outcomes.

Teacher's Corner Simple and Effective Confidence Intervals for Proportions and Differences of Proportions Result from Adding Two Successes and Two Failures

Abstract: Abstract The standard confidence intervals for proportions and their differences used in introductory statistics courses have poor performance, the actual coverage probability often being much lower than intended. However, simple adjustments of these intervals based on adding four pseudo observations, half of each type, perform surprisingly well even for small samples. To illustrate, for a broad variety of parameter settings with 10 observations in each sample, a nominal 95% interval for the difference of proportions has actual coverage probability below .93 in 88% of the cases with the standard interval but in only 1% with the adjusted interval; the mean distance between the nominal and actual coverage probabilities is .06 for the standard interval, but .01 for the adjusted one. In teaching with these adjusted intervals, one can bypass awkward sample size guidelines and use the same formulas with small and large samples.

Simple and Effective Confidence Intervals for Proportions and Differences of Proportions Result from Adding Two Successes and Two Failures

Abstract: The standard confidence intervals for proportions and their differences used in introductory statistics courses have poor performance, the actual coverage probability often being much lower than intended. However, simple adjustments of these intervals based on adding four pseudo observations, half of each type, perform surprisingly well even for small samples. To illustrate, for a broad variety of parameter settings with 10 observations in each sample, a nominal 95% interval for the difference of proportions has actual coverage probability below .93 in 88% of the cases with the standard interval but in only 1% with the adjusted interval; the mean distance between the nominal and actual coverage probabilities is .06 for the standard interval, but .01 for the adjusted one. In teaching with these adjusted intervals, one can bypass awkward sample size guidelines and use the same formulas with small and large samples.

Plasma DNA as a Prognostic Marker in Trauma Patients

Abstract: Background: Recently, much interest has developed in the potential use of plasma DNA as a diagnostic and monitoring tool. We hypothesized that plasma DNA is increased in patients with trauma and may be prognostic in such patients. Methods: We studied 84 patients who had sustained an acute blunt traumatic injury. We measured plasma DNA by a real-time quantitative PCR assay for the β-globin gene. Blood samples were collected at a median time of 60 min following injury. Blood samples were also obtained from 27 control subjects. Results: The median plasma DNA concentrations in the control, minor/moderate trauma (Injury Severity Score <16; n = 47), and major trauma (Injury Severity Score ≥16; n = 37) groups were 3154 kilogenome-equivalents/L, 13 818 kilogenome-equivalents/L, and 181 303 kilogenome-equivalents/L, respectively. Plasma DNA concentrations in patients with adverse outcomes, including acute lung injury, acute respiratory distress syndrome, and death, had 11.6- to 12-fold higher plasma DNA concentrations than those who did not develop these complications. At a cutoff of 232 719 kilogenome-equivalents/L, the sensitivities of plasma DNA analysis for the prediction of acute lung injury, acute respiratory distress syndrome, and death were 100% (95% confidence interval, 100–100%), 100% (95% confidence interval, 100–100%), and 78% (95% confidence interval, 40–97%), respectively. The respective specificities were 81% (95% confidence interval, 71–89%), 80% (95% confidence interval, 70–88%), and 82% (95% confidence interval, 71–90%). Conclusions: Plasma DNA is increased after trauma and may be a potentially valuable prognostic marker for these patients.

Systematic review and economic evaluation of Helicobacter pylori eradication treatment for non›ulcer dyspepsia

Abstract: Objectives: To evaluate efficacy and cost effectiveness of Helicobacter pylori eradication treatment in patients with non-ulcer dyspepsia infected with H pylori. Design: Systematic review of randomised controlled trials comparing H pylori eradication with placebo or another drug treatment. Results were incorporated into a Markov model comparing health service costs and benefits of H pylori eradication with antacid treatment over one year. Data sources: Six electronic databases were searched for randomised controlled trials from January 1966 to May 2000. Experts in the field, pharmaceutical companies, and journals were contacted for information on any unpublished trials. Trial reports were reviewed according to predefined eligibility and quality criteria. Main outcome measures: Relative risk reduction for remaining dyspeptic symptoms (the same or worse) at 3-12 months. Cost per dyspepsia-free month estimated from Markov model based on estimated relative risk reduction. Results: Twelve trials were included in the systematic review, nine of which evaluated dyspepsia at 3-12 months in 2541 patients. H pylori eradication treatment was significantly superior to placebo in treating non-ulcer dyspepsia (relative risk reduction 9% (95% confidence interval 4% to 14%)), one case of dyspepsia being cured for every 15 people treated. H pylori eradication cost £56 per dyspepsia-free month during first year after treatment. Conclusion:H pylori eradication may be cost effective treatment for non-ulcer dyspepsia in infected patients but further evidence is needed on decision makers9 willingness to pay for relief of dyspepsia.

Confidence interval criteria for assessment of dose proportionality

Abstract: Purpose. The aim of this work was a pragmatic, statistically sound and clinically relevant approach to dose-proportionality analyses that is compatible with common study designs.

Residential Radon Gas Exposure and Lung Cancer The Iowa Radon Lung Cancer Study

Abstract: Exposure to high concentrations of radon progeny (radon) produces lung cancer in both underground miners and experimentally exposed laboratory animals. To determine the risk posed by residential radon exposure, the authors performed a population-based, case-control epidemiologic study in Iowa from 1993 to 1997. Subjects were female Iowa residents who had occupied their current home for at least 20 years. A total of 413 lung cancer cases and 614 age-frequency-matched controls were included in the final analysis. Excess odds were calculated per 11 working-level months for exposures that occurred 5-19 years (WLM(5-19)) prior to diagnosis for cases or prior to time of interview for controls. Eleven WLM(5-19) is approximately equal to an average residential radon exposure of 4 pCl/liter (148 Bq/m3) during this period. After adjustment for age, smoking, and education, the authors found excess odds of 0.50 (95% confidence interval: 0.004, 1.81) and 0.83 (95% percent confidence interval: 0.11, 3.34) using categorical radon exposure estimates for all cases and for live cases, respectively. Slightly lower excess odds of 0.24 (95 percent confidence interval: -0.05, 0.92) and 0.49 (95 percent confidence interval: 0.03, 1.84) per 11 WLM(5-19) were noted for continuous radon exposure estimates for all subjects and live subjects only. The observed risk estimates suggest that cumulative ambient radon exposure presents an important environmental health hazard.

A multilevel model framework for meta-analysis of clinical trials with binary outcomes

Abstract: In this paper we explore the potential of multilevel models for meta-analysis of trials with binary outcomes for both summary data, such as log-odds ratios, and individual patient data. Conventional fixed effect and random effects models are put into a multilevel model framework, which provides maximum likelihood or restricted maximum likelihood estimation. To exemplify the methods, we use the results from 22 trials to prevent respiratory tract infections; we also make comparisons with a second example data set comprising fewer trials. Within summary data methods, confidence intervals for the overall treatment effect and for the between-trial variance may be derived from likelihood based methods or a parametric bootstrap as well as from Wald methods; the bootstrap intervals are preferred because they relax the assumptions required by the other two methods. When modelling individual patient data, a bias corrected bootstrap may be used to provide unbiased estimation and correctly located confidence intervals; this method is particularly valuable for the between-trial variance. The trial effects may be modelled as either fixed or random within individual data models, and we discuss the corresponding assumptions and implications. If random trial effects are used, the covariance between these and the random treatment effects should be included; the resulting model is equivalent to a bivariate approach to meta-analysis. Having implemented these techniques, the flexibility of multilevel modelling may be exploited in facilitating extensions to standard meta-analysis methods.