Effect measures in prevalence studies

There is still considerable confusion and debate about the appropriate methods for analyzing prevalence studies, and a number of recent papers have argued that prevalence ratios are the preferred method and that prevalence odds ratios should not be used. These arguments assert that the prevalence ratio is obviously the better measure and the odds ratio is "unintelligible." They have often been accompanied by demonstrations that when a disease is common the prevalence ratio and the prevalence odds ratio may differ substantially. However, this does not tell us which measure is the more valid to use. In fact, the prevalence odds ratio a) estimates the incidence rate ratio with fewer assumptions than are required for the prevalence ratio; b) can be estimated using the same methods as for the odds ratio in case-control studies, namely, the Mantel-Haenszel method and logistic regression; and c) provides practical, analytical, and theoretical consistency between analyses of a prevalence study and prevalence case-control analyses based on the same study population. For these reasons, the prevalence odds ratio will continue to be one of the standard methods for analyzing prevalence studies and prevalence case-control studies.     

应用Meta分析有效估计人群归因危险度百分比的探讨

人群归因危险度百分比 (populationattributableriskproportion ,PARP)是总体人群中某种疾病归因于某种因素的暴露所引起的发病 (死亡 )占全部发病 (死亡 )的百分比 ,反映该因素所引起的发病 (死亡 )占全部发病 (死亡 )的比重。通过PARP可了解各危险因素对人群中某疾病的发病所产生的影响 ,亦即消除某危险因素后 ,所产生的对预防该疾病的效果将占有多大比重。它能够为卫生政策的制订提供依据 ,有着重要的公共卫生的实际意义。目前常用的估计PARP的方法有两种 :一种是利用全国人群抽样调查获得的人群总暴露率来估计 ;另一种是利用某地区…     

Efficiency loss from categorizing quantitative exposures into qualitative exposures in case-control studies.

In the analysis of data from case-control studies, quantitative exposure variables are frequently categorized into qualitative exposure variables, such as quarters. The qualitative exposure variables may be scalar variables that take the median values of each quantile interval, or they may be vectors of indicator variables that represent each quantile interval. In a qualitative analysis, the scalar variables may be used to test the dose-response relation, while the indicator variables may be used to estimate odds ratios for each higher quantile interval versus the lowest. Qualitative analysis, implicitly and explicitly documented by many epidemiologists and biostatisticians, has several desirable advantages (including simple interpretation and robustness in the presence of a misspecified model or outlier values). In a quantitative analysis, the quantitative exposure variables may be directly regressed to test the dose-response relation, as well as to estimate odds ratios of interest. As this paper demonstrates, quantitative analysis is generally more efficient than qualitative analysis. Through a Monte Carlo simulation study, the authors estimated the loss of efficiency that results from categorizing a quantitative exposure variable by quartiles in case-control studies with a total of 200 cases and 200 controls. In the analysis of the dose-response relation, this loss is about 30% or more; the percentage may reach about 50% when the odds ratio for the fourth quartile interval versus the lowest is around 4. In estimating odds ratios, the loss of efficiency for the second, third, and fourth quartile intervals versus the lowest is around 90%, 75%, and 40%, respectively. The authors consider the pros and cons of each analytic approach, and they recommend that 1) qualitative analysis be used initially to estimate the odds ratios for each higher quantile interval versus the lowest to examine the dose-response relation and determine the appropriateness of the assumed underlying model; and 2) quantitative analysis be used to test the dose-response relation under a plausible log odds ratio model.     

The problem of profound mismeasurement and the power of epidemiological studies of diet and cancer

Ecological associations of fat intake with breast cancer incidence have not, in general, been corroborated by individual-based epidemiological studies. Profound mismeasurement, which, in these studies, probably typifies measures of dietary exposures in general and of fat in particular may, in part, explain this lack of agreement. To demonstrate the way in which error masks effects, we studied the impact of extreme mismeasurement in analysis of strong or moderate underlying associations using computer-simulated, case-control studies (300 cases, 300 controls). Severe error causes the mean and median odds ratios to be biased toward unity, tests for trend and upper category odds ratios to be often not significant, and lower category odds ratios frequently to exceed higher exposure ones. Important risk relationships can be concealed, despite careful design and analysis if there is substantial mismeasurement of exposure.     

Case-crossover analyses of air pollution exposure data: referent selection strategies and their implications for bias

The case-crossover design has been widely used to study the association between short-term air pollution exposure and the risk of an acute adverse health event. The design uses cases only; for each individual case, exposure just before the event is compared with exposure at other control (or "referent") times. Time-invariant confounders are controlled by making within-subject comparisons. Even more important in the air pollution setting is that time-varying confounders can also be controlled by design by matching referents to the index time. The referent selection strategy is important for reasons in addition to control of confounding. The case-crossover design makes the implicit assumption that there is no trend in exposure across the referent times. In addition, the statistical method that is used-conditional logistic regression-is unbiased only with certain referent strategies. We review here the case-crossover literature in the air pollution context, focusing on key issues regarding referent selection. We conclude with a set of recommendations for choosing a referent strategy with air pollution exposure data. Specifically, we advocate the time-stratified approach to referent selection because it ensures unbiased conditional logistic regression estimates, avoids bias resulting from time trend in the exposure series, and can be tailored to match on specific time-varying confounders.     

Indirect validation of a retrospective method of exposure assessment used in a nested case-control study of lung cancer and silica exposure.

Validations of retrospective methods of assessment used in occupational epidemiological studies have rarely been published. This study is an indirect validation of a quantitative retrospective assessment of exposure to silica used in a nested case-control study of lung cancer among workers at 29 metal mines and pottery factories in China. Indices of cumulative total dust and cumulative respirable dust were calculated by merging work histories with the historical exposure profile for each subject. To validate indirectly the methods of exposure assessment used in the study of lung cancer, trends for exposure response relation between the two indices of exposure to silica and risk of silicosis were evaluated with 376 patients with silicosis from the study population as the cases, and 1262 controls without silicosis for comparison. Age adjusted odds ratios (ORs) as a measure of risk of silicosis showed striking trends with both indices of exposure to silica. For cumulative respirable dust, the OR (95% confidence interval) rose from 7.6 (5.1-11.4) for low exposure to 20.0 (13.2-30.6) for medium exposure, and to 51.7 (31.0-86.8) for high exposure. The strength of the association between exposure to silica and risk of silicosis suggests that the retrospective assessment of exposure used in the case-control study of lung cancer would accurately reflect an exposure response relation between silica and lung cancer, if it existed.     

Comparison of self-reported lifetime sun exposure with two methods of cutaneous microtopography

There is currently no "gold standard" for measuring lifetime sun exposure. Exploration of alternatives to self-reports is important for examining illnesses related to ultraviolet light exposure. Using skin replicas obtained from 184 controls in a breast cancer case-control study (Toronto, Ontario, Canada, 2004-2005), the authors compared self-reported indicators of lifetime sun exposure with two measures of cutaneous microtopography, the Beagley-Gibson system and skin line counts. With the Beagley-Gibson system, significantly increased odds ratios were found for age (odds ratio (OR) = 1.10, 95% confidence interval (CI): 1.05, 1.16), spending 7 days outside per week during the summer (OR = 3.33, 95% CI: 1.48, 7.50), and lifetime number of sunlamp sessions. Significantly decreased odds ratios were found for having darker skin, ever giving birth, and ever using sunlamps. With the skin line count approach, significant positive associations were found for age (OR = 2.31, 95% CI: 1.23, 4.35), age squared, duration of working in outdoor jobs (OR = 0.88, 95% CI: 0.79, 0.98), and average number of outdoor activities per week at ages 20-29 years (OR = 1.05, 95% CI: 1.00, 1.10). While the Beagley-Gibson method was associated with more variables than the skin line count method, both methods require further refinement before graded skin replicas can be recommended as a substitute for self-report measures.     

Use of twins as mutual proxy respondents in a case-control study of breast cancer: effect of item nonresponse and misclassification

A case-control study of breast cancer in twins diagnosed before 1988 was used to characterize the effects on odds ratios when proxy responses from co-twins are used. North American disease-discordant pairs were ascertained through advertisements, and mailed questionnaires were returned from both members of 671 pairs and from one member of 391 pairs. Biases from the proxy response were attributed to nonresponse or misclassification. Nonresponse varied according to type of exposure variable, depth of detail requested, joint exposure status of the pair, respondent's case-control status, zygosity, and social closeness of the pair. Misclassification was minimal, generally nondifferential, and a high degree of reliability between the proxy and self-report was indicated by the kappa statistic and the intraclass correlation coefficient. By using double-respondent pairs, a method was developed to adjust proxy responses for both sources of bias. These adjustments resulted in minor changes to the odds ratios for the variables studied (age at menarche, reproductive factors, and hormone use). A larger difference was observed between the odds ratios based on all pairs and those based on double-respondent pairs only. These findings demonstrate that, for these variables in this population, twins are reliable proxies for each other and that results from single-respondent pairs should be included.     

Patterns of bias due to differential misclassification by case–control status in a case–control study

BACKGROUND: Case-control study is still one of the most commonly used study designs in epidemiological research. Misclassification of case-control status remains a significant issue because it will bias the results of a case-control study. There exist two types of misclassification, differential versus nondifferential. It is commonly accepted that nondifferential misclassification will bias the results of the study towards the null hypothesis. Conversely, no reports have assessed the impact and direction of differential misclassification on odds ratio (OR) estimate. The goal of the present study is to demonstrate by statistical derivation that patterns exist on the bias induced by differential misclassification. METHODS: Based on a 2 x 2 case-control study design, we derive the odds ratio without misclassification, and those with misclassification according to: (1) controls are misclassified as cases by exposure status; (2) cases are misclassified as controls by exposure status; and (3) both controls and cases are misclassified by exposure status simultaneously. Furthermore, mathematical derivations are shown for each of the ratios of the two odds ratios with and without misclassification. These methods are carried out by simulation analyses. RESULTS: Simulation analyses show that quite a number of biased odds ratios tend to move away from the null hypothesis and result in approaching zero or infinity with increasing proportion of misclassification among cases, controls, or both. These patterns are associated with the exposure status and the values of unbiased odds ratio (<1, 1, or >1). CONCLUSIONS: Our findings suggest that, unlike nondifferential misclassification, differential misclassification of case-control status in a case-control study may not weaken the exposure-outcome association towarding the null hypothesis. Care needs to be taken for interpreting the results of a case-control study when there exists differential misclassification bias, a practical issue in epidemiological research.     

On the use of population attributable fraction to determine sample size for case-control studies of gene-environment interaction

Most methods for calculating the sample size needed to detect gene-environment interactions use odds ratios to measure the effect size. We show that for any combination of susceptible genotype prevalence and exposure prevalence and their associated risks, the odds ratio measuring strength of interaction corresponds to a population attributable fraction (PAF) because of interaction and vice versa. Simultaneous consideration of odds ratio for interaction and the associated PAF attributable to interaction provides additional insight to investigators evaluating the feasibility and public health relevance of a proposed study. We considered gene-environment interactions on a multiplicative scale, and assumed a dichotomous environmental exposure variable and a single two-allele disease-susceptibility locus. Our results show, for example, that for studies of exposures and genotypes that are common in a population (30%-50%), the PAF for interaction is large (>27%) even if the odds ratio for interaction is only moderate (approximately 2). If simultaneous estimates of interaction odds ratio and PAF indicate that the PAF is so large as to be implausible, the investigator may decide to reevaluate the study design based on detecting a more reasonable PAF. In this case, the associated odds ratio for interaction will be weaker and a considerably larger sample size may be needed.     

Occupational exposure to power frequency magnetic fields and risk of non-Hodgkin lymphoma

Objectives

To investigate the risk of non‐Hodgkin lymphoma (NHL) using a job‐exposure matrix (JEM) to assess exposure to occupational magnetic fields at the power frequencies of 50/60 Hz.

Methods

The study population consisted of 694 cases of NHL, first diagnosed between 1 January 2000 and 31 August 2001, and 694 controls from two regions in Australia, matched by age, sex and region of residence. A detailed occupational history was given by each subject. Exposure to power frequency magnetic fields was estimated using a population‐based JEM which was specifically developed in the United States to assess occupational magnetic field exposure. The cumulative exposure distribution was divided into quartiles and adjusted odds ratios were calculated using the lowest quartile as the referent group.

Results

For the total work history, the odds ratio (OR) for workers in the upper quartile of exposure was 1.48 (95% CI 1.02 to 2.16) compared to the referent (p value for trend was 0.006). When the exposure was lagged by 5 years the OR was 1.59 (95% CI 1.07 to 2.36) (p value for trend was 0.003). Adjusting for other occupational exposures did not significantly alter the results.

Conclusions

These findings provide weak support for the hypothesis that occupational exposure to 50/60 Hz magnetic fields increases the risk of NHL.     

Mortality from nephritis and nephrosis in the fibreglass manufacturing industry

OBJECTIVES: To investigate the question of whether there is an association between exposure to silica or respirable glass fibre and mortality from nephritis or nephrosis among workers in fibrous glass wool manufacturing facilities. METHODS: A case-control study with cases and controls derived from the Owens Corning mortality surveillance system. Two case-control analyses were carried out, one where the cases are defined with nephritis or nephrosis as the underlying cause of death and one where cases are defined as those where nephritis or nephrosis is either the underlying or a contributing cause of death. RESULTS: There is no consistent relation between respirable fibres or respirable silica and nephritis or nephrosis when the analysis is based either on underlying cause only or on underlying plus contributing cause of death. None of the sociodemographic variables considered suggests an increased risk when considering both underlying and contributing cause of death. CONCLUSIONS: These data would seem to support the contention that the most accurate picture of renal disease will be gained from the use of all information on the death certificate and not only the underlying cause. For these data, all odds ratios (ORs) for respirable fibres and silica based on both underlying and contributing cause of death are < 1 with the exception of the highest exposure to silica which is slightly > 1 (OR = 1.04). Although these results do not prove that there is no association between nephritis and nephrosis and exposure to fibreglass or silica in the fibreglass manufacturing environment, they do not support the assertion that such an association exists.

 

     

Exposure to beryllium and occurrence of lung cancer: a reexamination of findings from a nested case-control study

OBJECTIVE: Our aim was to reanalyze a nested case-control study of beryllium and lung cancer because we identified analysis and study design issues that could have led to the elevated odds ratios obtained in the study. METHODS: We reanalyzed the data using nontransformed exposure metrics instead of log-transformed metrics used in the publication. We identified and examined effects on estimated odds ratios of imbalances between cases and controls caused by the control selection method. RESULTS: This reanalysis found no elevated odds ratios for any exposure variable. CONCLUSION:: Our conclusions differ from the authors' interpretation that the findings are due to a causal relationship between beryllium exposure and lung cancer. Our alternative explanation is that they may be due to methodological problems that could have been controlled by closer matching of controls to cases. CLINICAL SIGNIFICANCE: This study challenges conclusions made from a large case-control study concerning beryllium-lung cancer associations. Occupational medicine practitioners may want to integrate findings from this study into advice they give beryllium-exposed workers concerned about lung cancer.     

Comparison of asbestos exposure assessments by next-of-kin respondents, by an occupational hygienist, and by a job-exposure matrix from the National Occupational Hazard Survey

BACKGROUND: Assessments of occupational exposures in case-control studies of rapidly fatal illnesses often rely on data from next-of-kin respondents, which may be inaccurate. METHODS: Three methods for assessing exposure to asbestos from case-control data on mesothelioma, including next-of-kin assessment, expert assessment, and use of a generic job-exposure matrix (JEM). Interview data [Spirtas et al. (1994): Occup Environ Med 51:804-811] were reviewed to determine exposure status by an occupational hygienist (C.R.) who was unaware of disease status. Exposure odds ratios were calculated using standard methods, and measures of agreement included the kappa statistic and conditional and marginal odds ratios. RESULTS: Expert assessment detected higher proportions of exposed subjects than the next-of-kin respondents or JEM methods. The disease-exposure odds ratios were highest for respondents, perhaps because of recall bias, and lowest for the JEM method. The agreement was highest between the respondent and expert assessments. A combination of respondent's assessment and JEM assessment led to the best prediction of the expert's assessment. Results for spouse respondents were similar to those for other "next-of-kin" respondents. CONCLUSIONS: Expert assessments were the most plausible, but the data indicate that disease associations could also be detected with the other exposure assessment methods. Using some combination of the proxy respondent's assessment and the JEM assessment, one can predict the expert's assessment. A strategy that relied on the respondent's assessment when it was positive and otherwise obtained an expert assessment could reduce costs with little error, compared to expert assessment on all subjects.     

Behavior of the exposure odds ratio in a case-control study when the hazard function is not constant over time

In case-control studies of adverse drug effects there is rarely much evidence to support the assumption that the hazard function among users is constant during therapy. Nonetheless, this assumption is often implicitly made. We will use both clinical reasoning and a simple model to show how non-constancy of the hazard function affects odds ratio interpretation. When the hazard function is non-constant and there is more than one temporal pattern of drug usage, the odds ratio will estimate a weighted mean of incidence ratios with weights, dependent on the corresponding fractions of person-time. If the duration-specific incidence ratios differ widely, the odds ratio will depend not only on the drug but also on its usage pattern in the study population. This may explain some of the large regional odds ratio variation for dipyrone-related agranulocytosis in the International Agranulocytosis and Aplastic Anemia Study (JAMA 1986; 256: 1749–1757 [9].     

Combining matched and unmatched control groups in case-control studies

Multiple control groups in case-control studies are used to control for different sources of confounding. For example, cases can be contrasted with matched controls to adjust for multiple genetic or unknown lifestyle factors and simultaneously contrasted with an unmatched population-based control group. Inclusion of different control groups for a single exposure analysis yields several estimates of the odds ratio, all using only part of the data. Here the authors introduce an easy way to combine odds ratios from several case-control analyses with the same cases. The approach is based upon methods used for meta-analysis but takes into account the fact that the same cases are used and that the estimated odds ratios are therefore correlated. Two ways of estimating this correlation are discussed: sandwich methodology and the bootstrap. Confidence intervals for the pooled estimates and a test for checking whether the odds ratios in the separate case-control studies differ significantly are derived. The performance of the method is studied by simulation and by applying the methods to a large study on risk factors for thrombosis, the MEGA Study (1999-2004), wherein cases with first venous thrombosis were included with a matched control group of partners and an unmatched population-based control group.     

Potential misinterpretation of the case-only study to assess gene-environment interaction.

Novel epidemiologic study designs are often required to assess gene-environment interaction. A design using only cases, without controls, is one of several approaches that have been proposed as more efficient alternatives to the typical random sampling of cases and controls. However, it has not been pointed out that a case-only analysis estimates a different interaction parameter than does a traditional case-control analysis: The latter typically estimates departure from multiplicative population odds or rate ratios, depending on the method of control selection, while the former estimates departure from multiplicative risk ratios if genotype and environmental exposure are not associated in the population. These parameters are approximately equal if the disease risk is small at all levels of the study variables. The authors quantify the impact of allowing for higher disease risk among gene carriers, a relevant situation when the gene under study is highly penetrant. Their findings show that the cross-product ratio computed from case-only data may be substantially smaller than the odds ratio computed from case-control data and may therefore underestimate either the population odds or the rate ratio. Thus, to avoid misinterpretation of interaction parameters estimated from case-only data, the definition of multiplicative interaction should be made explicit.     

Nested case-control study of occupational chemical exposures and prostate cancer in aerospace and radiation workers

BACKGROUND: To date, little is known about the potential contributions of occupational exposure to chemicals to the etiology of prostate cancer. Previous studies examining associations suffered from limitations including the reliance on mortality data and inadequate exposure assessment. METHODS: We conducted a nested case-control study of 362 cases and 1,805 matched controls to examine the association between occupational chemical exposures and prostate cancer incidence. Workers were employed between 1950 and 1992 at a nuclear energy and rocket engine-testing facility in Southern California. We obtained cancer-incidence data from the California Cancer Registry and seven other state cancer registries. Data from company records were used to construct a job exposure matrix (JEM) for occupational exposures to hydrazine, trichloroethylene (TCE), polycyclic aromatic hydrocarbons (PAHs), benzene and mineral oil. Associations between chemical exposures and prostate cancer incidence were assessed in conditional logistic regression models. RESULTS: With adjustment for occupational confounders, including socioeconomic status, occupational physical activity, and exposure to the other chemicals evaluated, the odds ratio for low/moderate TCE exposure was 1.3; 95%CI = 0.8 to 2.1, and for high TCE exposure was 2.1; 95%CI = 1.2 to 3.9. Furthermore, we noted a positive trend between increasing levels of TCE exposure and prostate cancer (P-value for trend = 0.02). CONCLUSION: Our results suggest that high levels of TCE exposure are associated with prostate cancer among workers in our study population.     

不同研究设计的相对危险度估计

根据不同的研究设计，在队列研究中可分别估计累积发病率比（ＣＩＲ）、发病密度比（ＩＤＲ）和发病比值（ＩＯＲ）。当疾病罕见时，这３个比之值很接近。在病例对照研究中，是否可用暴露比值比（ＥＯＲ）作为ＣＩＲ或ＩＤＲ的无偏估计，取决于抽样方法和疾病率高低。本文还讨论了在不同抽样设计下ＥＯＲ与ＣＩＲ或ＩＤＲ之间的关系。