Abstract Clinic-based cohort studies enroll patients on first being admitted to the clinic, and follow them as part of usual care, with interest being in the marginal mean of the outcome process. As the required frequency of follow-up varies among patients, these studies often feature irregular visit times, with no two patients sharing a visit time. Inverse-intensity weighting has been developed to handle this, however it requires that the visit process be conditionally independent of the outcome given the observed history. When patients schedule visits in response to changes in their health (for example a disease flare), the conditional independence assumption is no longer plausible, leading to biased results. We suggest additional information that can be collected to ensure that conditional independence holds, and examine how this might be used in the analysis. This allows clinic-based cohort studies to be used to determine longitudinal outcomes without incurring bias due to irregular follow-up.
{"title":"Meeting the Assumptions of Inverse-Intensity Weighting for Longitudinal Data Subject to Irregular Follow-Up: Suggestions for the Design and Analysis of Clinic-Based Cohort Studies","authors":"E. Pullenayegum","doi":"10.1515/em-2018-0016","DOIUrl":"https://doi.org/10.1515/em-2018-0016","url":null,"abstract":"Abstract Clinic-based cohort studies enroll patients on first being admitted to the clinic, and follow them as part of usual care, with interest being in the marginal mean of the outcome process. As the required frequency of follow-up varies among patients, these studies often feature irregular visit times, with no two patients sharing a visit time. Inverse-intensity weighting has been developed to handle this, however it requires that the visit process be conditionally independent of the outcome given the observed history. When patients schedule visits in response to changes in their health (for example a disease flare), the conditional independence assumption is no longer plausible, leading to biased results. We suggest additional information that can be collected to ensure that conditional independence holds, and examine how this might be used in the analysis. This allows clinic-based cohort studies to be used to determine longitudinal outcomes without incurring bias due to irregular follow-up.","PeriodicalId":37999,"journal":{"name":"Epidemiologic Methods","volume":"1 1","pages":""},"PeriodicalIF":0.0,"publicationDate":"2020-01-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"91146608","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Abstract Objectives The methods to estimate the population attributable risk (PAR) of a single risk factor or the combined PAR of multiple risk factors have been extensively studied and well developed. Ideally, the estimation of combined PAR of multiple risk factors should be based on large cohort studies, which account for both the joint distributions of risk exposures and for their interactions. However, because such individual-level data are often lacking, many studies estimate the combined PAR using a comparative risk assessment framework. It involves estimating PAR of each risk factor based on its prevalence and relative risk, and then combining the individual PARs using an approach that relies on two key assumptions: that the distributions of exposures to the risk factors are independent and that the relative risks are multiplicative. While such assumptions rarely hold true in practice, no studies have investigated the magnitude of bias incurred if the assumptions are violated. Methods Using simulation-based models, we compared the combined PARs obtained with this approach to the more accurate estimates of PARs that are available when the joint distributions of exposures and risks can be established. Results We show that the assumptions of exposure independence and risk multiplicativity are sufficient but not necessary for the combined PAR to be unbiased. In the simplest situation of two risk factors, the bias of this approach is a function of the strength of association and the magnitude of risk interaction, for any values of exposure prevalence and their associated risks. In some cases, the combined PAR can be strongly under- or over-estimated, even if the two assumptions are only slightly violated. Conclusions We encourage researchers to quantify likely biases in their use of the M–S method, and here, we provided level plots and R code to assist.
{"title":"A comparison of approaches for estimating combined population attributable risks (PARs) for multiple risk factors","authors":"Y. Ruan, S. Walter, C. Friedenreich, D. Brenner","doi":"10.1515/em-2019-0021","DOIUrl":"https://doi.org/10.1515/em-2019-0021","url":null,"abstract":"Abstract Objectives The methods to estimate the population attributable risk (PAR) of a single risk factor or the combined PAR of multiple risk factors have been extensively studied and well developed. Ideally, the estimation of combined PAR of multiple risk factors should be based on large cohort studies, which account for both the joint distributions of risk exposures and for their interactions. However, because such individual-level data are often lacking, many studies estimate the combined PAR using a comparative risk assessment framework. It involves estimating PAR of each risk factor based on its prevalence and relative risk, and then combining the individual PARs using an approach that relies on two key assumptions: that the distributions of exposures to the risk factors are independent and that the relative risks are multiplicative. While such assumptions rarely hold true in practice, no studies have investigated the magnitude of bias incurred if the assumptions are violated. Methods Using simulation-based models, we compared the combined PARs obtained with this approach to the more accurate estimates of PARs that are available when the joint distributions of exposures and risks can be established. Results We show that the assumptions of exposure independence and risk multiplicativity are sufficient but not necessary for the combined PAR to be unbiased. In the simplest situation of two risk factors, the bias of this approach is a function of the strength of association and the magnitude of risk interaction, for any values of exposure prevalence and their associated risks. In some cases, the combined PAR can be strongly under- or over-estimated, even if the two assumptions are only slightly violated. Conclusions We encourage researchers to quantify likely biases in their use of the M–S method, and here, we provided level plots and R code to assist.","PeriodicalId":37999,"journal":{"name":"Epidemiologic Methods","volume":"83 1","pages":""},"PeriodicalIF":0.0,"publicationDate":"2020-01-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"84028539","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Anna S. Frank, D. Matteson, H. Solvang, A. Lupattelli, H. Nordeng
Abstract This manuscript extends the definition of the Absolute Standardized Mean Difference (ASMD) for binary exposure (M = 2) to cases for M > 2 on multiple imputed data sets. The Maximal Maximized Standardized Difference (MMSD) and the Maximal Averaged Standardized Difference (MASD) were proposed. For different percentages, missing data were introduced in covariates in the simulated data based on the missing at random (MAR) assumption. We then investigate the performance of these two metric definitions using simulated data of full and imputed data sets. The performance of the MASD and the MMSD were validated by relating the balance metrics to estimation bias. The results show that there is an association between the balance metrics and bias. The proposed balance diagnostics seem therefore appropriate to assess balance for the generalized propensity score (GPS) under multiple imputation.
本文将二元暴露(M = 2)的绝对标准化平均差(ASMD)的定义扩展到多个输入数据集上M > 2的情况。提出了最大最大标准化差(MMSD)和最大平均标准化差(MASD)。基于随机缺失(missing at random, MAR)假设，在模拟数据的协变量中引入不同百分比的缺失数据。然后，我们使用完整和输入数据集的模拟数据来研究这两种度量定义的性能。通过将平衡度量与估计偏差相关联，验证了MASD和MMSD的性能。结果表明，在平衡指标和偏差之间存在关联。因此，所提出的平衡诊断似乎适合于评估多重归算下广义倾向评分(GPS)的平衡。
{"title":"Extending balance assessment for the generalized propensity score under multiple imputation","authors":"Anna S. Frank, D. Matteson, H. Solvang, A. Lupattelli, H. Nordeng","doi":"10.1515/em-2019-0003","DOIUrl":"https://doi.org/10.1515/em-2019-0003","url":null,"abstract":"Abstract This manuscript extends the definition of the Absolute Standardized Mean Difference (ASMD) for binary exposure (M = 2) to cases for M > 2 on multiple imputed data sets. The Maximal Maximized Standardized Difference (MMSD) and the Maximal Averaged Standardized Difference (MASD) were proposed. For different percentages, missing data were introduced in covariates in the simulated data based on the missing at random (MAR) assumption. We then investigate the performance of these two metric definitions using simulated data of full and imputed data sets. The performance of the MASD and the MMSD were validated by relating the balance metrics to estimation bias. The results show that there is an association between the balance metrics and bias. The proposed balance diagnostics seem therefore appropriate to assess balance for the generalized propensity score (GPS) under multiple imputation.","PeriodicalId":37999,"journal":{"name":"Epidemiologic Methods","volume":"2 1","pages":""},"PeriodicalIF":0.0,"publicationDate":"2020-01-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"80279658","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
E. B. Rose, J. Roy, R. Castillo-Neyra, M. Ross, C. Condori-Pino, J. Peterson, César Náquira-Velarde, M. Levy
Abstract Objectives Containing domestic vector infestation requires the ability to swiftly locate and treat infested homes. In urban settings where vectors are heterogeneously distributed throughout a dense housing matrix, the task of locating infestations can be challenging. Here, we present a novel stochastic compartmental model developed to help locate infested homes in urban areas. We designed the model using infestation data for the Chagas disease vector species Triatoma infestans in Arequipa, Peru. Methods Our approach incorporates disease vector counts at each observed house, and the vector’s complex spatial dispersal dynamics. We used a Bayesian method to augment the observed data, estimate the insect population growth and dispersal parameters, and determine posterior infestation probabilities of households. We investigated the properties of the model through simulation studies, followed by field testing in Arequipa. Results Simulation studies showed the model to be accurate in its estimates of two parameters of interest: the growth rate of a domestic triatomine bug colony and the probability of a triatomine bug successfully invading a new home after dispersing from an infested home. When testing the model in the field, data collection using model estimates was hindered by low household participation rates, which severely limited the algorithm and in turn, the model’s predictive power. Conclusions While future optimization efforts must improve the model’s capabilities when household participation is low, our approach is nonetheless an important step toward integrating data with predictive modeling to carry out evidence-based vector surveillance in cities.
{"title":"A real-time search strategy for finding urban disease vector infestations","authors":"E. B. Rose, J. Roy, R. Castillo-Neyra, M. Ross, C. Condori-Pino, J. Peterson, César Náquira-Velarde, M. Levy","doi":"10.1515/em-2020-0001","DOIUrl":"https://doi.org/10.1515/em-2020-0001","url":null,"abstract":"Abstract Objectives Containing domestic vector infestation requires the ability to swiftly locate and treat infested homes. In urban settings where vectors are heterogeneously distributed throughout a dense housing matrix, the task of locating infestations can be challenging. Here, we present a novel stochastic compartmental model developed to help locate infested homes in urban areas. We designed the model using infestation data for the Chagas disease vector species Triatoma infestans in Arequipa, Peru. Methods Our approach incorporates disease vector counts at each observed house, and the vector’s complex spatial dispersal dynamics. We used a Bayesian method to augment the observed data, estimate the insect population growth and dispersal parameters, and determine posterior infestation probabilities of households. We investigated the properties of the model through simulation studies, followed by field testing in Arequipa. Results Simulation studies showed the model to be accurate in its estimates of two parameters of interest: the growth rate of a domestic triatomine bug colony and the probability of a triatomine bug successfully invading a new home after dispersing from an infested home. When testing the model in the field, data collection using model estimates was hindered by low household participation rates, which severely limited the algorithm and in turn, the model’s predictive power. Conclusions While future optimization efforts must improve the model’s capabilities when household participation is low, our approach is nonetheless an important step toward integrating data with predictive modeling to carry out evidence-based vector surveillance in cities.","PeriodicalId":37999,"journal":{"name":"Epidemiologic Methods","volume":"1 1","pages":""},"PeriodicalIF":0.0,"publicationDate":"2020-01-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"89387382","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
{"title":"The mean prevalence","authors":"F. Habibzadeh, P. Habibzadeh","doi":"10.1515/em-2019-0033","DOIUrl":"https://doi.org/10.1515/em-2019-0033","url":null,"abstract":"","PeriodicalId":37999,"journal":{"name":"Epidemiologic Methods","volume":"15 1","pages":""},"PeriodicalIF":0.0,"publicationDate":"2020-01-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"81165656","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
J. Ferguson, Fabrizio Maturo, S. Yusuf, M. O’Donnell
Abstract When estimating population attributable fractions (PAF), it is common to partition a naturally continuous exposure into a categorical risk factor. While prior risk factor categorization can help estimation and interpretation, it can result in underestimation of the disease burden attributable to the exposure as well as biased comparisons across different exposures and risk factors. Here, we propose sensible PAF estimands for continuous exposures under a potential outcomes framework. In contrast to previous approaches, we incorporate estimation of the minimum risk exposure value (MREV) into our procedures. While for exposures such as tobacco usage, a sensible value of the MREV is known, often it is unknown and needs to be estimated. Second, in the setting that the MREV value is an extreme-value of the exposure lying in the distributional tail, we argue that the natural estimator of PAF may be both statistically biased and highly volatile; instead, we consider a family of modified PAFs which include the natural estimate of PAF as a limit. A graphical comparison of this set of modified PAF for differing risk factors may be a better way to rank risk factors as intervention targets, compared to the standard PAF calculation. Finally, we analyse the bias that may ensue from prior risk factor categorization, examining whether categorization is ever a good idea, and suggest interpretations of categorized-estimands within a causal inference setting.
{"title":"Population attributable fractions for continuously distributed exposures","authors":"J. Ferguson, Fabrizio Maturo, S. Yusuf, M. O’Donnell","doi":"10.1515/em-2019-0037","DOIUrl":"https://doi.org/10.1515/em-2019-0037","url":null,"abstract":"Abstract When estimating population attributable fractions (PAF), it is common to partition a naturally continuous exposure into a categorical risk factor. While prior risk factor categorization can help estimation and interpretation, it can result in underestimation of the disease burden attributable to the exposure as well as biased comparisons across different exposures and risk factors. Here, we propose sensible PAF estimands for continuous exposures under a potential outcomes framework. In contrast to previous approaches, we incorporate estimation of the minimum risk exposure value (MREV) into our procedures. While for exposures such as tobacco usage, a sensible value of the MREV is known, often it is unknown and needs to be estimated. Second, in the setting that the MREV value is an extreme-value of the exposure lying in the distributional tail, we argue that the natural estimator of PAF may be both statistically biased and highly volatile; instead, we consider a family of modified PAFs which include the natural estimate of PAF as a limit. A graphical comparison of this set of modified PAF for differing risk factors may be a better way to rank risk factors as intervention targets, compared to the standard PAF calculation. Finally, we analyse the bias that may ensue from prior risk factor categorization, examining whether categorization is ever a good idea, and suggest interpretations of categorized-estimands within a causal inference setting.","PeriodicalId":37999,"journal":{"name":"Epidemiologic Methods","volume":"232 1","pages":""},"PeriodicalIF":0.0,"publicationDate":"2020-01-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"75758425","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
R. Lyles, S. Cunningham, Suprateek Kundu, Q. Bassat, I. Mandomando, C. Sacoor, Victor Akelo, D. Onyango, Emily Zielinski-Gutierrez, Allan W. Taylor
Abstract Objectives The Child Health and Mortality Prevention Surveillance (CHAMPS) Network is designed to elucidate and track causes of under-5 child mortality and stillbirth in multiple sites in sub-Saharan Africa and South Asia using advanced surveillance, laboratory and pathology methods. Expert panels provide an arguable gold standard determination of underlying cause of death (CoD) on a subset of child deaths, in part through examining tissue obtained via minimally invasive tissue sampling (MITS) procedures. We consider estimating a population-level distribution of CoDs based on this sparse but precise data, in conjunction with data on subgrouping characteristics that are measured on the broader population of cases and are potentially associated with selection for MITS and with cause-specific mortality. Methods We illustrate how estimation of each underlying CoD proportion using all available data can be addressed equivalently in terms of a Horvitz-Thompson adjustment or a direct standardization, uncovering insights relevant to the designation of appropriate subgroups to adjust for non-representative sampling. Taking advantage of the functional form of the result when expressed as a multinomial distribution-based maximum likelihood estimator, we propose small-sample adjustments to Bayesian credible intervals based on Jeffreys or related weakly informative Dirichlet prior distributions. Results Our analyses of early data from CHAMPS sites in Kenya and Mozambique and accompanying simulation studies demonstrate the validity of the adjustment approach under attendant assumptions, together with marked performance improvements associated with the proposed adjusted Bayesian credible intervals. Conclusions Adjustment for non-representative sampling of those validated via gold standard diagnostic methods is a critical endeavor for epidemiologic studies like CHAMPS that seek extrapolation of CoD proportion estimates.
{"title":"Extrapolating sparse gold standard cause of death designations to characterize broader catchment areas","authors":"R. Lyles, S. Cunningham, Suprateek Kundu, Q. Bassat, I. Mandomando, C. Sacoor, Victor Akelo, D. Onyango, Emily Zielinski-Gutierrez, Allan W. Taylor","doi":"10.1515/em-2019-0031","DOIUrl":"https://doi.org/10.1515/em-2019-0031","url":null,"abstract":"Abstract Objectives The Child Health and Mortality Prevention Surveillance (CHAMPS) Network is designed to elucidate and track causes of under-5 child mortality and stillbirth in multiple sites in sub-Saharan Africa and South Asia using advanced surveillance, laboratory and pathology methods. Expert panels provide an arguable gold standard determination of underlying cause of death (CoD) on a subset of child deaths, in part through examining tissue obtained via minimally invasive tissue sampling (MITS) procedures. We consider estimating a population-level distribution of CoDs based on this sparse but precise data, in conjunction with data on subgrouping characteristics that are measured on the broader population of cases and are potentially associated with selection for MITS and with cause-specific mortality. Methods We illustrate how estimation of each underlying CoD proportion using all available data can be addressed equivalently in terms of a Horvitz-Thompson adjustment or a direct standardization, uncovering insights relevant to the designation of appropriate subgroups to adjust for non-representative sampling. Taking advantage of the functional form of the result when expressed as a multinomial distribution-based maximum likelihood estimator, we propose small-sample adjustments to Bayesian credible intervals based on Jeffreys or related weakly informative Dirichlet prior distributions. Results Our analyses of early data from CHAMPS sites in Kenya and Mozambique and accompanying simulation studies demonstrate the validity of the adjustment approach under attendant assumptions, together with marked performance improvements associated with the proposed adjusted Bayesian credible intervals. Conclusions Adjustment for non-representative sampling of those validated via gold standard diagnostic methods is a critical endeavor for epidemiologic studies like CHAMPS that seek extrapolation of CoD proportion estimates.","PeriodicalId":37999,"journal":{"name":"Epidemiologic Methods","volume":"19 1","pages":""},"PeriodicalIF":0.0,"publicationDate":"2020-01-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"90912114","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
W. W. Loh, B. Moerkerke, T. Loeys, S. Vansteelandt
Abstract Decomposing an exposure effect on an outcome into separate natural indirect effects through multiple mediators requires strict assumptions, such as correctly postulating the causal structure of the mediators, and no unmeasured confounding among the mediators. In contrast, interventional indirect effects for multiple mediators can be identified even when – as often – the mediators either have an unknown causal structure, or share unmeasured common causes, or both. Existing estimation methods for interventional indirect effects require calculating each distinct indirect effect in turn. This can quickly become unwieldy or unfeasible, especially when investigating indirect effect measures that may be modified by observed baseline characteristics. In this article, we introduce simplified estimation procedures for such heterogeneous interventional indirect effects using interventional effect models. Interventional effect models are a class of marginal structural models that encode the interventional indirect effects as causal model parameters, thus readily permitting effect modification by baseline covariates using (statistical) interaction terms. The mediators and outcome can be continuous or noncontinuous. We propose two estimation procedures: one using inverse weighting by the counterfactual mediator density or mass functions, and another using Monte Carlo integration. The former has the advantage of not requiring an outcome model, but is susceptible to finite sample biases due to highly variable weights. The latter has the advantage of consistent estimation under a correctly specified (parametric) outcome model, but is susceptible to biases due to extrapolation. The estimators are illustrated using publicly available data assessing whether the indirect effects of self-efficacy on fatigue via self-reported post-traumatic stress disorder symptoms vary across different levels of negative coping among health care workers during the COVID-19 outbreak.
{"title":"Heterogeneous indirect effects for multiple mediators using interventional effect models","authors":"W. W. Loh, B. Moerkerke, T. Loeys, S. Vansteelandt","doi":"10.1515/em-2020-0023","DOIUrl":"https://doi.org/10.1515/em-2020-0023","url":null,"abstract":"Abstract Decomposing an exposure effect on an outcome into separate natural indirect effects through multiple mediators requires strict assumptions, such as correctly postulating the causal structure of the mediators, and no unmeasured confounding among the mediators. In contrast, interventional indirect effects for multiple mediators can be identified even when – as often – the mediators either have an unknown causal structure, or share unmeasured common causes, or both. Existing estimation methods for interventional indirect effects require calculating each distinct indirect effect in turn. This can quickly become unwieldy or unfeasible, especially when investigating indirect effect measures that may be modified by observed baseline characteristics. In this article, we introduce simplified estimation procedures for such heterogeneous interventional indirect effects using interventional effect models. Interventional effect models are a class of marginal structural models that encode the interventional indirect effects as causal model parameters, thus readily permitting effect modification by baseline covariates using (statistical) interaction terms. The mediators and outcome can be continuous or noncontinuous. We propose two estimation procedures: one using inverse weighting by the counterfactual mediator density or mass functions, and another using Monte Carlo integration. The former has the advantage of not requiring an outcome model, but is susceptible to finite sample biases due to highly variable weights. The latter has the advantage of consistent estimation under a correctly specified (parametric) outcome model, but is susceptible to biases due to extrapolation. The estimators are illustrated using publicly available data assessing whether the indirect effects of self-efficacy on fatigue via self-reported post-traumatic stress disorder symptoms vary across different levels of negative coping among health care workers during the COVID-19 outbreak.","PeriodicalId":37999,"journal":{"name":"Epidemiologic Methods","volume":"28 1","pages":""},"PeriodicalIF":0.0,"publicationDate":"2020-01-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"77992838","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Abstract The study of dementia risk factors is complicated by the competing risk of dying. The standard approaches are the cause-specific Cox proportional hazard model with deaths treated as censoring events (and removed from the risk set) and the Fine and Gray sub-distribution hazard model in which those who die remain in the risk set. An alternative approach is to modify the risk set between these extremes. We propose a novel method of doing this based on estimating the time at which the person might have been diagnosed if they had not died using a parametric survival model, and then applying the cause-specific and Fine and Gray models to the modified dataset. We compare these methods using data on dementia from the Australian Longitudinal Study on Women’s Health and discuss the assumptions and limitations of each model. The results from survival models to assess risk factors for dementia varied considerably between the cause-specific model and the models designed to account for competing risks. Therefore, when assessing risk factors in the presence of competing risks it is important to examine results from: the cause-specific model, different models which account for competing risks, and the model which assesses risk factors associated with the competing risk.
死亡的竞争风险使痴呆危险因素的研究变得复杂。标准方法是病因特异性Cox比例风险模型，其中死亡被视为审查事件(并从风险集中删除)，以及Fine和Gray子分布风险模型，其中死亡的人仍在风险集中。另一种方法是修改这两个极端之间的风险设置。我们提出了一种新的方法，该方法基于使用参数生存模型估计如果患者没有死亡，则患者可能被诊断的时间，然后将原因特定模型和Fine and Gray模型应用于修改后的数据集。我们使用澳大利亚妇女健康纵向研究的痴呆数据对这些方法进行比较，并讨论每个模型的假设和局限性。用于评估痴呆风险因素的生存模型的结果在病因特异性模型和用于考虑竞争风险的模型之间差异很大。因此，在评估存在竞争风险的风险因素时，重要的是要检查以下结果:特定原因模型，考虑竞争风险的不同模型，以及评估与竞争风险相关的风险因素的模型。
{"title":"A comparison of cause-specific and competing risk models to assess risk factors for dementia","authors":"M. Waller, G. Mishra, A. Dobson","doi":"10.1515/em-2019-0036","DOIUrl":"https://doi.org/10.1515/em-2019-0036","url":null,"abstract":"Abstract The study of dementia risk factors is complicated by the competing risk of dying. The standard approaches are the cause-specific Cox proportional hazard model with deaths treated as censoring events (and removed from the risk set) and the Fine and Gray sub-distribution hazard model in which those who die remain in the risk set. An alternative approach is to modify the risk set between these extremes. We propose a novel method of doing this based on estimating the time at which the person might have been diagnosed if they had not died using a parametric survival model, and then applying the cause-specific and Fine and Gray models to the modified dataset. We compare these methods using data on dementia from the Australian Longitudinal Study on Women’s Health and discuss the assumptions and limitations of each model. The results from survival models to assess risk factors for dementia varied considerably between the cause-specific model and the models designed to account for competing risks. Therefore, when assessing risk factors in the presence of competing risks it is important to examine results from: the cause-specific model, different models which account for competing risks, and the model which assesses risk factors associated with the competing risk.","PeriodicalId":37999,"journal":{"name":"Epidemiologic Methods","volume":"36 1","pages":""},"PeriodicalIF":0.0,"publicationDate":"2020-01-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"89724042","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
Pub Date : 2019-12-01Epub Date: 2019-08-02DOI: 10.1515/em-2018-0011
Emily J Huang, Ravi Varadhan, Michelle C Carlson
In studies of clinical phenotypes, such as dementia, disability, and frailty, participants are typically assessed at in-person clinic visits. Thus, the precise time of onset for the phenotype is unknown. The discreteness of the clinic visits yields grouped event time data. We investigate how to perform a risk factor analysis in the case of grouped data. Since visits can be months to years apart, numbers of ties can be large, causing the exact tie-handling method of the Cox model to be computationally infeasible. We propose two, new, computationally efficient approximations to the exact method: Laplace approximation and an analytic approximation. Through extensive simulation studies, we compare these new methods to the Prentice-Gloeckler model and the Cox model using Efron's and Breslow's tie-handling methods. In addition, we compare the methods in an application to a large cohort study (N = 3,605) on the development of clinical frailty in older adults. In our simulations, the Laplace approximation has low bias in all settings, and the analytic approximation has low bias in settings where the regression coefficient is not large in magnitude. Their corresponding confidence intervals also have approximately the nominal coverage probability. In the data application, the results from the approximations are nearly identical to that of the Prentice-Gloeckler model.
{"title":"Modeling of Clinical Phenotypes Assessed at Discrete Study Visits.","authors":"Emily J Huang, Ravi Varadhan, Michelle C Carlson","doi":"10.1515/em-2018-0011","DOIUrl":"10.1515/em-2018-0011","url":null,"abstract":"<p><p>In studies of clinical phenotypes, such as dementia, disability, and frailty, participants are typically assessed at in-person clinic visits. Thus, the precise time of onset for the phenotype is unknown. The discreteness of the clinic visits yields grouped event time data. We investigate how to perform a risk factor analysis in the case of grouped data. Since visits can be months to years apart, numbers of ties can be large, causing the exact tie-handling method of the Cox model to be computationally infeasible. We propose two, new, computationally efficient approximations to the exact method: Laplace approximation and an analytic approximation. Through extensive simulation studies, we compare these new methods to the Prentice-Gloeckler model and the Cox model using Efron's and Breslow's tie-handling methods. In addition, we compare the methods in an application to a large cohort study (<i>N</i> = 3,605) on the development of clinical frailty in older adults. In our simulations, the Laplace approximation has low bias in all settings, and the analytic approximation has low bias in settings where the regression coefficient is not large in magnitude. Their corresponding confidence intervals also have approximately the nominal coverage probability. In the data application, the results from the approximations are nearly identical to that of the Prentice-Gloeckler model.</p>","PeriodicalId":37999,"journal":{"name":"Epidemiologic Methods","volume":"8 1","pages":""},"PeriodicalIF":0.0,"publicationDate":"2019-12-01","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"https://www.ncbi.nlm.nih.gov/pmc/articles/PMC7806193/pdf/nihms-1618534.pdf","citationCount":null,"resultStr":null,"platform":"Semanticscholar","paperid":"38821278","PeriodicalName":null,"FirstCategoryId":null,"ListUrlMain":null,"RegionNum":0,"RegionCategory":"","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":"OA","EPubDate":null,"PubModel":null,"JCR":null,"JCRName":null,"Score":null,"Total":0}
京公网安备 11010802042870号
京ICP备2023020795号-1
扫码关注我们
求助内容：
应助结果提醒方式：