Design and conduct of occupational epidemiology studies: III. Design aspects of case-control studies

Currently available approaches for the design of occupational case-control studies are reviewed. An accompanying paper reviews methods of analysis. We commence by drawing a distinction between cohort-based and registry-based studies. Methods for selecting cases and controls are then reviewed, including cumulative incidence and incidence density sampling, matching, sources of controls, and issues in control selection. Finally, the advantages and disadvantages of the case-control approach are summarized.     

Occupational risk factors in Alzheimer's disease: a review assessing the quality of published epidemiological studies

Epidemiological evidence of an association between Alzheimer's disease (AD) and the most frequently studied occupational exposures--pesticides, solvents, electromagnetic fields (EMF), lead and aluminium--is inconsistent. Epidemiological studies published up to June of 2003 were systematically searched through PubMed and Toxline. Twenty-four studies (21 case-control and 3 cohort studies) were included. Median GQI was 36.6% (range 19.5-62.9%). Most of the case-control studies had a GQI of <50%. The study with the highest score was a cohort study. Likelihood of exposure misclassification bias affected 18 of the 24 studies. Opportunity for bias arising from the use of surrogate informants affected 17 studies, followed by disease misclassification (11 studies) and selection bias (10 studies). Eleven studies explored the relationship of AD with solvents, seven with EMF, six with pesticides, six with lead and three with aluminium. For pesticides, studies of greater quality and prospective design found increased and statistically significant associations. For the remaining occupational agents, the evidence of association is less consistent (for solvents and EMF) or absent (for lead and aluminium).     

Advancing epidemiologic studies of occupational injury—Approaches and future directions

Despite increased attention to the epidemiology of occupational injuries and work-related musculoskeletal disorders, little is known about the relative contribution of risk factors for specific injuries and diseases and effective ways to prevent them. Building on the knowledge and ideas presented in the previous papers in this special issue, this paper describes the salient issues regarding the needs and the opportunities in design, conduct, and analysis of epidemiologic research on occupational injury. There is a clear need for standardization of definitions on exposure variables and health outcomes, since basic terminology is not used consistently across studies. Improved surveillance systems with linkage to other databases offer excellent opportunities to evaluate trends in risk factors and occupational injuries. New opportunities for etiologic studies include the case-crossover design, which focuses on relations between intermittent exposures and injuries, as well as several hybrid designs that combine advantages and strength of the traditional case-control and cohort study designs. Various suggestions are presented to capture the dynamics of exposure, including both intermittent and continuous factors in the assessment strategy. The importance of cohort studies is emphasized to clarify the temporal relations between causes and effects in work-related injury research. Finally, intervention studies are much needed and a hierarchical approach is suggested, starting with observational surveys, continuing with quasi-experimental studies and concluding with randomized controlled experiments of those interventions that hold the largest promise. Am. J. Ind. Med. 32:180–183, 1997. © 1997 Wiley-Liss Inc.     

2001-2010年我国伤害预防病例对照研究及队列研究文献质量评价

目的评价2001-2010年我国公开发表的伤害预防病例对照研究和队列研究文献的质量。方法文献来源于四个主要的中文数据库和两个英文数据库。采用Newcastle-Ottawa-Scale（NOS）工具分别对纳入的伤害预防病例对照研究和队列研究文献进行方法学质量评价。结果 2001-2010年我国伤害预防病例对照研究和队列研究文献分别有84篇和13篇。病例对照研究中,职业人员和儿童青少年分别占45.2%和36.9%,伤害类型中,多种伤害占71.4%。队列研究中,儿童青少年占46.2%,伤害类型中,多种伤害占92.3%。病例对照研究的方法学质量得分均在5分及以上,平均（6.1±0.7）分;但研究在＂暴露因素的确定＂、＂无应答率＂以及＂病例的确定是否恰当＂条目上得分较低。队列研究的方法学质量得分以6分和7分为主,平均（6.3±0.9）分;但研究在＂结局指标的评价＂条目上得分较低。结论我国的伤害病例对照研究和队列研究数量较少,研究人群以儿童青少年和职业人员为主,伤害类型以多种伤害为主,质量相对较高。     

Effect of cigarette smoking in epidemiological studies of lung cancer

This paper describes a method for adjusting the analysis of occupational/environmental lung cancer risks for the effects of cigarette smoking in cohort and case-control studies. The method uses a function that relates an individual's death rate to his age and cigarette smoking history. Two such functions are examined. The first depends on total packs of cigarettes smoked and age. The second, based on the multistage theory of carcinogenesis, depends on age, age at start of smoking, and subsequent smoking rates. The lung cancer rates predicted by these two functions are compared to those observed in cohort studies of male British physicians and U.S. veterans, and in a case-control study of non-Hispanic white men in New Mexico. Neither of the cohort data sets distinguished the fit of the two functions. The New Mexico data were fit better by the second function, though both functions overpredicted death rates among ex-smokers. Each function explained substantially more variation in the New Mexico data than did any of several logistic regression models involving categorical variables for age and smoking.     

Occupational factors and pancreatic cancer

The relation between occupational factors and pancreatic cancer has been studied by two different approaches: a population based case-control study with two series of controls and a retrospective cohort study based on register data. With both approaches, some support was found for an association with occupational exposure to petroleum products. Associations were also indicated with exposure to paint thinner (case-control study) and work in painting and in paint and varnish factories (cohort study), for exposure to detergents, floor cleaning agents, or polish (case-control study) and with floor polishing or window cleaning (cohort study), and for exposure to refuse (case-control study) and work in refuse disposal plants (cohort study).     

Occupational factors and pancreatic cancer.

The relation between occupational factors and pancreatic cancer has been studied by two different approaches: a population based case-control study with two series of controls and a retrospective cohort study based on register data. With both approaches, some support was found for an association with occupational exposure to petroleum products. Associations were also indicated with exposure to paint thinner (case-control study) and work in painting and in paint and varnish factories (cohort study), for exposure to detergents, floor cleaning agents, or polish (case-control study) and with floor polishing or window cleaning (cohort study), and for exposure to refuse (case-control study) and work in refuse disposal plants (cohort study).     

Bias in studies of parental self-reported occupational exposure and childhood cancer

Several case-control studies have demonstrated positive associations between parental occupational exposures and childhood cancer. However, an overestimation of risk estimates due to recall bias is of concern. The magnitude and nature of this bias were explored using data from a German case-control study on childhood leukemia conducted between 1992 and 1997. A moderate overreporting of occupational exposures by fathers was observed, particularly for the prenatal period. Overreporting was most apparent when the time between exposure and interview was short. It was also found that job titles were no satisfactory substitute for information on specific occupational exposures. The results of this analysis emphasize the need for more sophisticated exposure assessment methods in epidemiologic studies of childhood cancer. However, because future case-control studies will at least partially rely on questionnaire data, improvements including probing questions, better interview techniques, and validation studies are indicated.     

A plea to stop using the case-control design in retrospective database studies

The case-control design is widely used in retrospective database studies, often leading to spectacular findings. However, results of these studies often cannot be replicated, and the advantage of this design over others is questionable. To demonstrate the shortcomings of applications of this design, we replicate two published case-control studies. The first investigates isotretinoin and ulcerative colitis using a simple case-control design. The second focuses on dipeptidyl peptidase-4 inhibitors and acute pancreatitis, using a nested case-control design. We include large sets of negative control exposures (where the true odds ratio is believed to be 1) in both studies. Both replication studies produce effect size estimates consistent with the original studies, but also generate estimates for the negative control exposures showing substantial residual bias. In contrast, applying a self-controlled design to answer the same questions using the same data reveals far less bias. Although the case-control design in general is not at fault, its application in retrospective database studies, where all exposure and covariate data for the entire cohort are available, is unnecessary, as other alternatives such as cohort and self-controlled designs are available. Moreover, by focusing on cases and controls it opens the door to inappropriate comparisons between exposure groups, leading to confounding for which the design has few options to adjust for. We argue that this design should no longer be used in these types of data. At the very least, negative control exposures should be used to prove that the concerns raised here do not apply.     

Epidemiology of occupational risks: methodological problems

The main methodological problems encountered in occupational epidemiology are related to the particularity of the risk factors concerned. It is necessary to account for the "Healthy Worker Effect" in the design, analysis and interpretation of surveys since the very presence of workers at particular jobs is related to their health. In cohort studies, it is necessary to have precise knowledge of the work environment corresponding to different jobs as exposure information is generally not individual, objective and continuously recorded. In case-control studies also, information on occupational exposures is a key issue, not only from the point of view of differential bias but also because exposure that is too loosely defined leads to dilution and reduced accuracy.     

Selection of controls in case-control studies. III. Design options.

Several design options available in the planning stage of case-control studies are examined. Topics covered include matching, control/case ratio, choice of nested case-control or case-cohort design, two-stage sampling, and other methods that can be used for control selection. The effect of potential problems in obtaining comparable accuracy of exposure is also examined. A discussion of the difficulty in meeting the principles of study base, deconfounding, and comparable accuracy (S. Wacholder et al. Am J Epidemiol 1992;135:1019-28) in a single study completes this series of papers.     

Analysis of case-cohort designs

The case-cohort design is most useful in analyzing time to failure in a large cohort in which failure is rare. Covariate information is collected from all failures and a representative sample of censored observations. Sampling is done without respect to time or disease status, and, therefore, the design is more flexible than a nested case-control design. Despite the efficiency of the methods, case-cohort designs are not often used because of perceived analytic complexity. In this article, we illustrate computation of a simple variance estimator and discuss model fitting techniques in SAS. Three different weighting methods are considered. Model fitting is demonstrated in an occupational exposure study of nickel refinery workers. The design is compared to a nested case-control design with respect to analysis and efficiency in a small simulation. In this example, case-cohort sampling from the full cohort was more efficient than using a comparable nested case-control design.     

Pregnancy outcome following maternal organic solvent exposure: A meta-analysis of epidemiologic studies

Background Evidence of fetal damage or demise from occupational organic solvent levels that are not toxic to the pregnant woman is inconsistent in the medical literature. The risk for major malformations and spontaneous abortion from maternal inhalation of organic solvent exposure during pregnancy was summarized using meta-analysis. Methods Medline, Toxline, and Dissertation Abstracts databases were searched to locate all research papers published in any language from 1966 to 1994. Included were studies that were case-control or cohort in design and indicated first trimester (or up to 20 weeks gestation for spontaneous abortion) maternal solvent exposure. A summary odds ratio (ORs) with 95% confidence intervals (CI) was calculated from research results combined by the Mantel-Haenszel method. Results In total, 559 studies were obtained from the literature search. Five studies for each outcome of interest qualified for inclusion in the analysis. The ORs for major malformations from five studies (n = 7,036 patients) was 1.64 (CI 1.16–2.30) and for spontaneous abortion from five studies (n = 2,899 patients) was 1.25 (CI 0.99–1.58). Conclusions Maternal occupational exposure to organic solvents is associated with a tendency toward an increased risk for spontaneous abortion and additional studies may affect the trend. There is a statistically significant association with major malformations which warrants further investigation. Am. J. Ind. Med. 34:288–292, 1998. © 1998 Wiley-Liss, Inc.     

Applicability of the simple independent action model to epidemiologic studies involving two factors and a dichotomous outcome

In epidemiologic case-control studies and occupational cohort studies involving more than one exposure, it is sometimes of interest to investigate the possibility that two exposures or factors have an effect that is mutually enhancing. This paper begins with a simple classic model for independence of effect and describes how this model can be applied to cohort and case-control studies. A ratio index, borrowed from the toxicologic literature, can be used to quantify departures from this null model for prospective cohort studies. Models additive in log nonresponse are appropriate in this context. Proper stratification will remove confounding effects, although the possibility that covarying susceptibilities among individuals in the population are masking or producing the appearance of synergy remains. However, under a generalized null model that requires simple independent action for each individual, but allows the response probabilities to vary among individuals, the population-based ratio parameter may not be one but should lie in a specified interval. In a case-control setting, the simple independent action model implies that the ratio of the bivariate exposure distribution for cases, divided by that for controls, should be additive in functions of the exposure levels, generalizing an earlier result. The index takes a different form when one of the factors is preventive rather than causal, and in this context, models additive in log risk become appropriate. An example is provided, and difficulties in interpretation are discussed.     

Bias in occupational epidemiology studies

The design of occupational epidemiology studies should be based on the need to minimise random and systematic error. The latter is the focus of this paper, and includes selection bias, information bias and confounding. Selection bias can be minimised by obtaining a high response rate (and by appropriate selection of the control group in a case-control study). In general, it is important to ensure that information bias is minimised and is also non-differential (for example, that the misclassification of exposure is not related to disease status) by collecting data in a standardised manner. A major concern in occupational epidemiology studies usually relates to confounding, because exposure has not been randomly allocated, and the groups under study may therefore have different baseline disease risks. For each of these types of bias, the goal should be to avoid the bias by appropriate study design and/or appropriate control in the analysis. However, it is also important to attempt to assess the likely direction and strength of biases that cannot be avoided or controlled.     

Analytic methods for two-stage case-control studies and other stratified designs

Nested case-control studies, or case-control studies within a cohort, combine the advantages of cohort studies with the efficiency of case-control studies. Case-control studies can often be viewed as having two stages; the first stage consists of vital status, disease, and basic covariate ascertainment, and the second stage consists of detailed covariate and exposure ascertainment. Breslow and Cain (1988) and Breslow and Zhao (1988) recently showed that conventional analyses of such two-stage studies may ignore some of the available information. In this paper, we show how one can adapt the pseudo-likelihood analyses developed by Kalbfleisch and Lawless (1988) to the analysis of data from two-stage case-control studies.     

Homocyst(e)ine and cardiovascular disease: a systematic review of the evidence with special emphasis on case-control studies and nested case-control studies

BACKGROUND: Elevated concentrations of homocyst(e)ine are thought to increase the risk of vascular diseases including coronary heart disease and cerebrovascular disease. METHODS: We searched MEDLINE (1966-1999), EMBASE (1974-1999), SciSearch (1974- 1999), and Dissertation Abstracts (1999) for articles and theses about homocyst(e)ine concentration and coronary heart disease and cerebrovascular disease. RESULTS: We included 57 publications (3 cohort studies, 12 nested case-control studies, 42 case-control studies) that reported results on 5518 people with coronary heart disease (11,068 control subjects) and 1817 people with cerebrovascular disease (4787 control subjects) in our analysis. For coronary heart disease, the summary odds ratios (OR) for a 5-micromol/l increase in homocyst(e)ine concentration were 1.06 (95% CI : 0.99-1.13) for 2 publications of cohort studies, 1.23 (95% CI : 1.07-1.41) for 10 publications of nested case-control studies, and 1.70 (95% CI : 1.50-1.93) for 26 publications of case-control studies. For cerebrovascular disease, the summary OR for a 5-micromol/l increase in homocyst(e)ine concentration were 1.10 (95% CI : 0.94-1.28) for 2 publications of cohort studies, 1.58 (95% CI : 1.35-1.85) for 5 publications of nested case-control studies, and 2.16 (95% CI : 1.65-2.82) for 17 publications of case-control studies. CONCLUSIONS: Prospective studies offer weaker support than case-control studies for an association between homocyst(e)ine concentration and cardiovascular disease. Although other lines of evidence support a role for homocyst(e)ine in the pathogenesis of cardiovascular disease, more information from prospective epidemiological studies or clinical trials is needed to clarify this role.     

Design and conduct of occupational epidemiology studies: II. Analysis of cohort data

This paper reviews strategies and statistical methods for analyzing data from occupational cohort studies. Emphasis is placed on the common methods for grouped data analysis involving external and internal comparison populations. Analysis procedures reviewed are standardized mortality ratio, standardized rate ratio, and Mantel-Haenszel techniques for estimating relative risks. Methods for control of confounding, assessment of effect modification, and allowance for disease latency are discussed. These concepts and procedures are illustrated with data from an historical cohort mortality study of workers from an asbestos textile plant.