Epidemiologic perspectives & innovations : EP+I最新文献

Causal inference requires an understanding of the conditions under which association equals causation. The exchangeability or no confounding assumption is well known and well understood as central to this task. More recently the epidemiologic literature has described additional assumptions related to the stability of causal effects. In this paper we extend the Sufficient Component Cause Model to represent one expression of this stability assumption--the Stable Unit Treatment Value Assumption. Approaching SUTVA from an SCC model helps clarify what SUTVA is and reinforces the connections between interaction and SUTVA.

The Integrated Health Interview Series (IHIS) is a public data repository that harmonizes four decades of the National Health Interview Survey (NHIS). The NHIS is the premier source of information on the health of the U.S. population. Since 1957 the survey has collected information on health behaviors, health conditions, and health care access. The long running time series of the NHIS is a powerful tool for health research. However, efforts to fully utilize its time span are obstructed by difficult documentation, unstable variable and coding definitions, and non-ignorable sample re-designs. To overcome these hurdles the IHIS, a freely available and web-accessible resource, provides harmonized NHIS data from 1969-2010. This paper describes the challenges of working with the NHIS and how the IHIS reduces such burdens. To demonstrate one potential use of the IHIS we examine utilization patterns in the U.S. from 1972-2008.

The past five years have seen a growth in the interest in systems approaches in epidemiologic research. These approaches may be particularly appropriate for social epidemiology. Social network analysis and agent-based models (ABMs) are two approaches that have been used in the epidemiologic literature. Social network analysis involves the characterization of social networks to yield inference about how network structures may influence risk exposures among those in the network. ABMs can promote population-level inference from explicitly programmed, micro-level rules in simulated populations over time and space. In this paper, we discuss the implementation of these models in social epidemiologic research, highlighting the strengths and weaknesses of each approach. Network analysis may be ideal for understanding social contagion, as well as the influences of social interaction on population health. However, network analysis requires network data, which may sacrifice generalizability, and causal inference from current network analytic methods is limited. ABMs are uniquely suited for the assessment of health determinants at multiple levels of influence that may couple with social interaction to produce population health. ABMs allow for the exploration of feedback and reciprocity between exposures and outcomes in the etiology of complex diseases. They may also provide the opportunity for counterfactual simulation. However, appropriate implementation of ABMs requires a balance between mechanistic rigor and model parsimony, and the precision of output from complex models is limited. Social network and agent-based approaches are promising in social epidemiology, but continued development of each approach is needed.

Background: In molecular epidemiology studies biospecimen data are collected, often with the purpose of evaluating the synergistic role between a biomarker and another feature on an outcome. Typically, biomarker data are collected on only a proportion of subjects eligible for study, leading to a missing data problem. Missing data methods, however, are not customarily incorporated into analyses. Instead, complete-case (CC) analyses are performed, which can result in biased and inefficient estimates.

Methods: Through simulations, we characterized the performance of CC methods when interaction effects are estimated. We also investigated whether standard multiple imputation (MI) could improve estimation over CC methods when the data are not missing at random (NMAR) and auxiliary information may or may not exist.

Results: CC analyses were shown to result in considerable bias and efficiency loss. While MI reduced bias and increased efficiency over CC methods under specific conditions, it too resulted in biased estimates depending on the strength of the auxiliary data available and the nature of the missingness. In particular, CC performed better than MI when extreme values of the covariate were more likely to be missing, while MI outperformed CC when missingness of the covariate related to both the covariate and outcome. MI always improved performance when strong auxiliary data were available. In a real study, MI estimates of interaction effects were attenuated relative to those from a CC approach.

Conclusions: Our findings suggest the importance of incorporating missing data methods into the analysis. If the data are MAR, standard MI is a reasonable method. Auxiliary variables may make this assumption more reasonable even if the data are NMAR. Under NMAR we emphasize caution when using standard MI and recommend it over CC only when strong auxiliary data are available. MI, with the missing data mechanism specified, is an alternative when the data are NMAR. In all cases, it is recommended to take advantage of MI's ability to account for the uncertainty of these assumptions.

This commentary intends to instigate discussions about epidemiologic estimates and their interpretation of attributable fractions (AFs) and the burden of disease (BOD) of cancers due to factors at workplaces. By examining recent work that aims to estimate the number of cancers attributable to shift-work in Britain, we suggest that (i) causal, (ii) practical and (iii) methodological areas of concern may deter us from attributable caseload estimations of cancers at this point in time. Regarding (i), such calculations may have to be avoided as long as we lack established causality between shift-work and the development of internal cancers. Regarding (ii), such calculations may have to be avoided as long as we can neither abandon shift-work nor identify personnel that may be unaffected by shift-work factors. Regarding (iii), there are at least four methodological pitfalls which are likely to make AF calculations uninterpretable at this stage. The four pitfalls are: (1) The use of Levin's 1953 formula in case of adjusted relative risks; (2) The use of broad definitions of exposure in calculations of AFs; (3) The non-additivity of AFs across different levels of exposure and covariables; (4) The fact that excess mortality counts are misleading due to the fact that a human being dies exactly once - a death may occur earlier or later, but a death cannot occur more than once nor can it be avoided altogether for any given individual. Overall, causal, practical and methodological areas of concern should be diligently considered when performing and interpreting AF or BOD computations which - at least at the present time - may not be defensible.

Adherence to a medical treatment means the extent to which a patient follows the instructions or recommendations by health professionals. There are direct and indirect ways to measure adherence which have been used for clinical management and research. Typically adherence measures are monitored over a long follow-up or treatment period, and some measurements may be missing due to death or other reasons. A natural question then is how to describe adherence behavior over the whole period in a simple way. In the literature, measurements over a period are usually combined just by using averages like percentages of compliant days or percentages of doses taken. In the paper we adapt an approach where patient adherence measures are seen as a stochastic process. Repeated measures are then analyzed as a Markov chain with finite number of states rather than as independent and identically distributed observations, and the transition probabilities between the states are assumed to fully describe the behavior of a patient. The patients can then be clustered or classified using their estimated transition probabilities. These natural clusters can be used to describe the adherence of the patients, to find predictors for adherence, and to predict the future events. The new approach is illustrated and shown to be useful with a simple analysis of a data set from the DART (Development of AntiRetroviral Therapy in Africa) trial in Uganda and Zimbabwe.

Most common diseases demonstrate familial aggregation; the ratio of the risk for relatives of affected people to the risk for relatives of unaffected people (the familial risk ratio)) > 1. This implies there are underlying genetic and/or environmental risk factors shared by relatives. The risk gradient across this underlying 'familial risk profile', which can be predicted from family history and measured familial risk factors, is typically strong. Under a multiplicative model, the ratio of the risk for people in the upper 25% of familial risk to the risk for those in the lower 25% (the inter-quartile risk gradient) is an order of magnitude greater than the familial risk ratio. If familial risk ratio = 2 for first-degree relatives, in terms of familial risk profile: (a) people in the upper quartile will be at more than 20 times the risk of those in the lower quartile; and (b) about 90% of disease will occur in people above the median. Historically, therefore, epidemiology has compared cases with controls dissimilar for underlying familial risk profile. Were gene-environment and gene-gene interactions to exist, environmental and genetic effects could be stronger for people with increased familial risk profile. Studies in which controls are better matched to cases for familial risk profile might be more informative, especially if both cases and controls are over-sampled for increased familial risk. Prospective family study cohort (ProF-SC) designs involving people across a range of familial risk profile provide such a resource for epidemiological, genetic, behavioural, psycho-social and health utilisation research. The prospective aspect gives credibility to risk estimates. The familial aspect allows family-based designs, matching for unmeasured factors, adjusting for underlying familial risk profile, and enhanced cohort maintenance.

Background: The WINPEPI computer programs for epidemiologists are designed for use in practice and research in the health field and as learning or teaching aids. The programs are free, and can be downloaded from the Internet. Numerous additions have been made in recent years.

Implementation: There are now seven WINPEPI programs: DESCRIBE, for use in descriptive epidemiology; COMPARE2, for use in comparisons of two independent groups or samples; PAIRSetc, for use in comparisons of paired and other matched observations; LOGISTIC, for logistic regression analysis; POISSON, for Poisson regression analysis; WHATIS, a "ready reckoner" utility program; and ETCETERA, for miscellaneous other procedures. The programs now contain 122 modules, each of which provides a number, sometimes a large number, of statistical procedures. The programs are accompanied by a Finder that indicates which modules are appropriate for different purposes. The manuals explain the uses, limitations and applicability of the procedures, and furnish formulae and references.

Conclusions: WINPEPI is a handy resource for a wide variety of statistical routines used by epidemiologists. Because of its ready availability, portability, ease of use, and versatility, WINPEPI has a considerable potential as a learning and teaching aid, both with respect to practical procedures in the planning and analysis of epidemiological studies, and with respect to important epidemiological concepts. It can also be used as an aid in the teaching of general basic statistics.

Background: Effectively responding to infectious disease outbreaks requires a well-informed response. Quantitative methods for analyzing outbreak data and estimating key parameters to characterize the spread of the outbreak, including the reproductive number and the serial interval, often assume that the data collected is complete. In reality reporting delays, undetected cases or lack of sensitive and specific tests to diagnose disease lead to reporting errors in the case counts. Here we provide insight on the impact that such reporting errors might have on the estimation of these key parameters.

Results: We show that when the proportion of cases reported is changing through the study period, the estimates of key epidemiological parameters are biased. Using data from the Influenza A/H1N1 outbreak in La Gloria, Mexico, we provide estimates of these parameters, accounting for possible reporting errors, and show that they can be biased by as much as 33%, if reporting issues are not accounted for.

Conclusions: Failure to account for missing data can lead to misleading and inaccurate estimates of epidemic parameters.

This commentary intends to instigate discussions about upcoming epidemiologic research, and its interpretation, into putative links between shift work, involving circadian disruption or chronodisruption [CD], and the development of internal cancers.In 2007, the International Agency for Research on Cancer (IARC) convened an expert group to examine the carcinogenicity of shift work, inter alia characterized by light exposures at unusual times. After a critical review of published data, the following was stated: "There is sufficient evidence in experimental animals for the carcinogenicity of light during the daily dark period (biological night)". However, in view of limited epidemiological evidence, it was overall concluded: "Shiftwork that involves circadian disruption is probably carcinogenic to humans (Group 2A)".Remarkably, the scenario around shift work, CD and internal cancers provides a unique case for "white-box" epidemiology: Research at many levels - from sub-cellular biochemistry, to whole cells, to organs, to organisms, including animals and humans - has suggested a series of quite precise and partly related causal mechanisms. This is in stark contrast to instances of "black box" or "stabs in the dark" epidemiology where causal mechanisms are neither known nor hypothesized or only poorly defined. The overriding theme that an adequate chronobiological organization of physiology can be critical for the protection against cancer builds the cornerstone of biological plausibility in this case.We can now benefit from biological plausibility in two ways: First, epidemiology should use biologically plausible insights into putative chains of causation between shift work and cancer to design future investigations. Second, when significant new data were to become available in coming years, IARC will re-evaluate cancer hazards associated with shift work. Biological plausibility may then be a key viewpoint to consider and, ultimately, to decide whether (or not) to pass from statistical associations, possibly detected in observational studies by then, to a verdict of causation.In the meantime, biological plausibility should not be invoked to facilitate publication of epidemiological research of inappropriate quality. Specific recommendations as to how to design, report and interpret epidemiological research into biologically plausible links between shift work and cancer are provided.Epidemiology is certainly a poor toolfor learning about the mechanismby which a disease is produced,but it has the tremendous advantagethat it focuses on the diseases and the deathsthat actually occur,and experience has shown that it continues to be second to none asa means of discovering linksin the chain of causationthat are capable of being broken.-Sir Richard Doll 1.