Kenneth Menglin Lee
Current Organizations / Institutes
The organizations/institutes from the scholar's latest article
Kenneth Menglin Lee has a total of 21 co-authors. Here are the top co-authors by number of shared publications:
Grace Meijuan Yang
Division of Supportive and Palliative Care, National Cancer Centre Singapore, Singapore, Singapore
3 shared publicationsYin Bun Cheung
Centre for Quantitative Medicine, Duke-NUS Medical School, Singapore, Singapore
3 shared publications- Yin Bun Cheung
Tampere University and Tampere University Hospital
3 shared publications Fan Li
Department of Biostatistics Yale School of Public Health New Haven CT USA
2 shared publications- Fan Li
Department of Biostatistics Yale School of Public Health New Haven Connecticut USA
2 shared publications Michael O. Harhay
Department of Biostatistics, Epidemiology and Informatics University of Pennsylvania Philadelphia PA USA
2 shared publications
Latest Publications
- Open Access
- 8 August 2026
- Published by SAGE Publications in Statistical Methods in Medical Research
- Vol. 35 (9) , 1980-2000
What is estimated in cluster randomized crossover trials with informative sizes? A survey of estimands and common estimators
The cluster randomized crossover (CRXO) trial, among other multi-period cluster randomized trial designs, can target average treatment effect (ATE) estimands that equally weigh the contributions of individuals (iATE), clusters (cATE), cluster-periods (cpATE), or periods (pATE). With these weighted ATE estimands, we define different forms of informative sizes, where the treatment effects vary according to cluster, period, and/or cluster-period sizes, causing these estimands to differ. Under such informative sizes, we survey which of the unweighted, inverse cluster-period size weighted, inverse cluster size weighted, and inverse period size weighted: (i) independence estimating equation, (ii) fixed effects model, (iii) exchangeable mixed effects model, and (iv) nested exchangeable mixed effects model treatment effect estimators are consistent for the aforementioned estimands in cross-sectional CRXO designs with continuous outcomes. We demonstrate that with informative sizes, the unweighted and weighted nested exchangeable mixed effects model estimators are not consistent for any meaningful estimand and can yield biased results. In contrast, the unweighted and weighted independence estimating equation, and under specific scenarios, the fixed effects model and exchangeable mixed effects model, can yield consistent and empirically unbiased estimators for meaningful estimands in CRXO trials.
- 8 August 2026
- Published by SAGE Publications in Statistical Methods in Medical Research
- Vol. 35 (9) , 1980-2000
Open AccessWhat is estimated in cluster randomized crossover trials with informative sizes? A survey of estimands and common estimators
The cluster randomized crossover (CRXO) trial, among other multi-period cluster randomized trial designs, can target average treatment effect (ATE) estimands that equally weigh the contributions of individuals (iATE), clusters (cATE), cluster-periods (cpATE), or periods (pATE). With these weighted ATE estimands, we define different forms of informative sizes, where the treatment effects vary according to cluster, period, and/or cluster-period sizes, causing these estimands to differ. Under such informative sizes, we survey which of the unweighted, inverse cluster-period size weighted, inverse cluster size weighted, and inverse period size weighted: (i) independence estimating equation, (ii) fixed effects model, (iii) exchangeable mixed effects model, and (iv) nested exchangeable mixed effects model treatment effect estimators are consistent for the aforementioned estimands in cross-sectional CRXO designs with continuous outcomes. We demonstrate that with informative sizes, the unweighted and weighted nested exchangeable mixed effects model estimators are not consistent for any meaningful estimand and can yield biased results. In contrast, the unweighted and weighted independence estimating equation, and under specific scenarios, the fixed effects model and exchangeable mixed effects model, can yield consistent and empirically unbiased estimators for meaningful estimands in CRXO trials.
- Open Access
Who's Winning? Clarifying Estimands Based on Win Statistics in Cluster Randomized Trials
Treatment effect estimands based on win statistics, including the win ratio, win odds, and win difference are increasingly popular targets for summarizing endpoints in clinical trials. Such win estimands offer an intuitive approach for prioritizing outcomes by clinical importance. The implementation and interpretation of win estimands is complicated in cluster randomized trials (CRTs), where researchers can target fundamentally different estimands on the individual-level or cluster-level. We numerically demonstrate that individual-pair and cluster-pair win estimands can substantially differ when cluster size is informative: where outcomes and/or treatment effects depend on cluster size. With such informative cluster sizes, individual-pair and cluster-pair win estimands can even yield opposite conclusions regarding treatment benefit. We describe consistent estimators for individual-pair and cluster-pair win estimands and propose a leave-one-cluster-out jackknife variance estimator for inference. Despite being consistent, our simulations highlight that some caution is needed when implementing individual-pair win estimators due to finite-sample bias. In contrast, cluster-pair win estimators are unbiased for their respective targets. Altogether, careful specification of the target estimand is essential when applying win estimators in CRTs. Failure to clearly define whether individual-pair or cluster-pair win estimands are of primary interest may result in answering a dramatically different question than intended.
Open AccessWho's Winning? Clarifying Estimands Based on Win Statistics in Cluster Randomized Trials
Treatment effect estimands based on win statistics, including the win ratio, win odds, and win difference are increasingly popular targets for summarizing endpoints in clinical trials. Such win estimands offer an intuitive approach for prioritizing outcomes by clinical importance. The implementation and interpretation of win estimands is complicated in cluster randomized trials (CRTs), where researchers can target fundamentally different estimands on the individual-level or cluster-level. We numerically demonstrate that individual-pair and cluster-pair win estimands can substantially differ when cluster size is informative: where outcomes and/or treatment effects depend on cluster size. With such informative cluster sizes, individual-pair and cluster-pair win estimands can even yield opposite conclusions regarding treatment benefit. We describe consistent estimators for individual-pair and cluster-pair win estimands and propose a leave-one-cluster-out jackknife variance estimator for inference. Despite being consistent, our simulations highlight that some caution is needed when implementing individual-pair win estimators due to finite-sample bias. In contrast, cluster-pair win estimators are unbiased for their respective targets. Altogether, careful specification of the target estimand is essential when applying win estimators in CRTs. Failure to clearly define whether individual-pair or cluster-pair win estimands are of primary interest may result in answering a dramatically different question than intended.
- Open Access
- 24 September 2025
- Published by Wiley in Statistics in Medicine
- Vol. 44 (20-22) , e70256
Analysis of Stepped‐Wedge Cluster Randomized Trials When Treatment Effects Vary by Exposure Time or Calendar Time
Stepped‐wedge cluster randomized trials (SW‐CRTs) are traditionally analyzed with models that assume an immediate and sustained treatment effect. Previous work has shown that making such an assumption in the analysis of SW‐CRTs when the true underlying treatment effect varies by exposure time can produce severely misleading estimates. Alternatively, the true underlying treatment effect might vary by calendar time. Comparatively less work has examined treatment effect structure misspecification in this setting. Here, we evaluate the behavior of the linear mixed effects model‐based immediate treatment effect, exposure time‐averaged treatment effect, and calendar time‐averaged treatment effect estimators in different scenarios where these estimators are misspecified for the true underlying treatment effect structure. We show that the immediate treatment effect estimator is relatively robust to bias when estimating a true underlying calendar time‐averaged treatment effect estimand. However, when there is a true underlying calendar (exposure) time‐varying treatment effect, misspecifying an analysis with an exposure (calendar) time‐averaged treatment effect estimator can yield severely misleading estimates which may converge to a value with the opposite sign of the true calendar (exposure) time‐averaged treatment effect estimand. In this article, we highlight these two different time scales on which treatment effects can vary in SW‐CRTs and clarify potential vulnerabilities that may arise when considering different types of time‐varying treatment effects in a SW design. Accordingly, we emphasize the need for researchers to carefully consider whether the treatment effect may vary as a function of exposure time or calendar time in the analysis of SW‐CRTs.
- 24 September 2025
- Published by Wiley in Statistics in Medicine
- Vol. 44 (20-22) , e70256
Open AccessAnalysis of Stepped‐Wedge Cluster Randomized Trials When Treatment Effects Vary by Exposure Time or Calendar Time
Stepped‐wedge cluster randomized trials (SW‐CRTs) are traditionally analyzed with models that assume an immediate and sustained treatment effect. Previous work has shown that making such an assumption in the analysis of SW‐CRTs when the true underlying treatment effect varies by exposure time can produce severely misleading estimates. Alternatively, the true underlying treatment effect might vary by calendar time. Comparatively less work has examined treatment effect structure misspecification in this setting. Here, we evaluate the behavior of the linear mixed effects model‐based immediate treatment effect, exposure time‐averaged treatment effect, and calendar time‐averaged treatment effect estimators in different scenarios where these estimators are misspecified for the true underlying treatment effect structure. We show that the immediate treatment effect estimator is relatively robust to bias when estimating a true underlying calendar time‐averaged treatment effect estimand. However, when there is a true underlying calendar (exposure) time‐varying treatment effect, misspecifying an analysis with an exposure (calendar) time‐averaged treatment effect estimator can yield severely misleading estimates which may converge to a value with the opposite sign of the true calendar (exposure) time‐averaged treatment effect estimand. In this article, we highlight these two different time scales on which treatment effects can vary in SW‐CRTs and clarify potential vulnerabilities that may arise when considering different types of time‐varying treatment effects in a SW design. Accordingly, we emphasize the need for researchers to carefully consider whether the treatment effect may vary as a function of exposure time or calendar time in the analysis of SW‐CRTs.
- Open Access
What is estimated in cluster randomized crossover trials with informative sizes? -- A survey of estimands and common estimators
In the analysis of cluster randomized trials (CRTs), previous work has defined two meaningful estimands: the individual-average treatment effect (iATE) and cluster-average treatment effect (cATE) estimand, to address individual and cluster-level hypotheses. In multi-period CRT designs, such as the cluster randomized crossover (CRXO) trial, additional weighted average treatment effect estimands help fully reflect the longitudinal nature of these trial designs, namely the cluster-period-average treatment effect (cpATE) and period-average treatment effect (pATE). We define different forms of informative sizes, where the treatment effects vary according to cluster, period, and/or cluster-period sizes, which subsequently cause these estimands to differ in magnitude. Under such conditions, we demonstrate which of the unweighted, inverse cluster-period size weighted, inverse cluster size weighted, and inverse period size weighted: (i.) independence estimating equation, (ii.) fixed effects model, (iii.) exchangeable mixed effects model, and (iv.) nested exchangeable mixed effects model treatment effect estimators are consistent for the aforementioned estimands in 2-period cross-sectional CRXO designs with continuous outcomes. We report a simulation study and conclude with a reanalysis of a CRXO trial testing different treatments on hospital length of stay among patients receiving invasive mechanical ventilation. Notably, with informative sizes, the unweighted and weighted nested exchangeable mixed effects model estimators are not consistent for any meaningful estimand and can yield biased results. In contrast, the unweighted and weighted independence estimating equation, and under specific scenarios, the fixed effects model and exchangeable mixed effects model, can yield consistent and empirically unbiased estimators for meaningful estimands in 2-period CRXO trials.
Open AccessWhat is estimated in cluster randomized crossover trials with informative sizes? -- A survey of estimands and common estimators
In the analysis of cluster randomized trials (CRTs), previous work has defined two meaningful estimands: the individual-average treatment effect (iATE) and cluster-average treatment effect (cATE) estimand, to address individual and cluster-level hypotheses. In multi-period CRT designs, such as the cluster randomized crossover (CRXO) trial, additional weighted average treatment effect estimands help fully reflect the longitudinal nature of these trial designs, namely the cluster-period-average treatment effect (cpATE) and period-average treatment effect (pATE). We define different forms of informative sizes, where the treatment effects vary according to cluster, period, and/or cluster-period sizes, which subsequently cause these estimands to differ in magnitude. Under such conditions, we demonstrate which of the unweighted, inverse cluster-period size weighted, inverse cluster size weighted, and inverse period size weighted: (i.) independence estimating equation, (ii.) fixed effects model, (iii.) exchangeable mixed effects model, and (iv.) nested exchangeable mixed effects model treatment effect estimators are consistent for the aforementioned estimands in 2-period cross-sectional CRXO designs with continuous outcomes. We report a simulation study and conclude with a reanalysis of a CRXO trial testing different treatments on hospital length of stay among patients receiving invasive mechanical ventilation. Notably, with informative sizes, the unweighted and weighted nested exchangeable mixed effects model estimators are not consistent for any meaningful estimand and can yield biased results. In contrast, the unweighted and weighted independence estimating equation, and under specific scenarios, the fixed effects model and exchangeable mixed effects model, can yield consistent and empirically unbiased estimators for meaningful estimands in 2-period CRXO trials.
- Open Access
- 29 April 2025
- Published by Wiley in Biometrical Journal
- Vol. 67 (3) , e70052
How Should Parallel Cluster Randomized Trials With a Baseline Period be Analyzed?—A Survey of Estimands and Common Estimators
The parallel cluster randomized trial with baseline (PB‐CRT) is a common variant of the standard parallel cluster randomized trial (P‐CRT). We define two natural estimands in the context of PB‐CRTs with informative cluster sizes, the individual‐average treatment effect (iATE) and cluster‐average treatment effect (cATE), to address individual and cluster‐level hypotheses. In this work, we theoretically derive the convergence of the unweighted and inverse cluster‐period size weighted (i) independence estimating equation (IEE), (ii) fixed‐effects (FE) model, (iii) exchangeable mixed‐effects (EME) model, and (iv) nested‐exchangeable mixed‐effects (NEME) model treatment effect estimators in a PB‐CRT with informative cluster sizes and continuous outcomes. Overall, we theoretically show that the unweighted and weighted IEE and FE models yield consistent estimators for the iATE and cATE estimands. Although mixed‐effects models yield inconsistent estimators to these two natural estimands under informative cluster sizes, we empirically demonstrate that the EME model is surprisingly robust to bias. This is in sharp contrast to the corresponding analyses in P‐CRTs and the NEME model in PB‐CRTs when informative cluster sizes are present, carrying implications for practice. We report a simulation study and conclude with a re‐analysis of a PB‐CRT examining the effects of community youth teams on improving mental health among adolescent girls in rural eastern India.
- 29 April 2025
- Published by Wiley in Biometrical Journal
- Vol. 67 (3) , e70052
Open AccessHow Should Parallel Cluster Randomized Trials With a Baseline Period be Analyzed?—A Survey of Estimands and Common Estimators
The parallel cluster randomized trial with baseline (PB‐CRT) is a common variant of the standard parallel cluster randomized trial (P‐CRT). We define two natural estimands in the context of PB‐CRTs with informative cluster sizes, the individual‐average treatment effect (iATE) and cluster‐average treatment effect (cATE), to address individual and cluster‐level hypotheses. In this work, we theoretically derive the convergence of the unweighted and inverse cluster‐period size weighted (i) independence estimating equation (IEE), (ii) fixed‐effects (FE) model, (iii) exchangeable mixed‐effects (EME) model, and (iv) nested‐exchangeable mixed‐effects (NEME) model treatment effect estimators in a PB‐CRT with informative cluster sizes and continuous outcomes. Overall, we theoretically show that the unweighted and weighted IEE and FE models yield consistent estimators for the iATE and cATE estimands. Although mixed‐effects models yield inconsistent estimators to these two natural estimands under informative cluster sizes, we empirically demonstrate that the EME model is surprisingly robust to bias. This is in sharp contrast to the corresponding analyses in P‐CRTs and the NEME model in PB‐CRTs when informative cluster sizes are present, carrying implications for practice. We report a simulation study and conclude with a re‐analysis of a PB‐CRT examining the effects of community youth teams on improving mental health among adolescent girls in rural eastern India.
- 4 February 2025
- Published by Oxford University Press (OUP) in Significance
- Vol. 22 (2) , 8-11
Advantage Federer: The expected value of tennis
Roger Federer said, “The best in the world are not the best because they win every point”. Kenneth Menglin Lee digs into point win percentage and considers how small differences can create huge sporting legends
- 4 February 2025
- Published by Oxford University Press (OUP) in Significance
- Vol. 22 (2) , 8-11
Advantage Federer: The expected value of tennis
Roger Federer said, “The best in the world are not the best because they win every point”. Kenneth Menglin Lee digs into point win percentage and considers how small differences can create huge sporting legends
- Open Access
- 19 November 2024
- Published by Springer Nature in BMC Medical Research Methodology
- Vol. 24 (1) , 1-2
Correction: Inclusion of unexposed clusters improves the precision of fixed effects analysis of stepped-wedge cluster randomized trials with binary and count outcomes
- 19 November 2024
- Published by Springer Nature in BMC Medical Research Methodology
- Vol. 24 (1) , 1-2
Open AccessCorrection: Inclusion of unexposed clusters improves the precision of fixed effects analysis of stepped-wedge cluster randomized trials with binary and count outcomes
- Open Access
- 28 October 2024
- Published by Springer Nature in BMC Medical Research Methodology
- Vol. 24 (1) , 1-11
Inclusion of unexposed clusters improves the precision of fixed effects analysis of stepped-wedge cluster randomized trials with binary and count outcomes
Background: The fixed effects model is a useful alternative to the mixed effects model for analyzing stepped-wedge cluster randomized trials (SW-CRTs). It controls for all time-invariant cluster-level confounders and has proper control of type I error when the number of clusters is small. While all clusters in a SW-CRT are typically designed to crossover from the control to receive the intervention, some trials can end with unexposed clusters (clusters that never receive the intervention), such as when a trial is terminated early due to safety concerns. It was previously unclear whether unexposed clusters would contribute to the estimation of the intervention effect in a fixed effects analysis. However, recent work has demonstrated that including an unexposed cluster can improve the precision of the intervention effect estimator in a fixed effects analysis of SW-CRTs with continuous outcomes. Still, SW-CRTs are commonly designed with binary outcomes and it is unknown if those previous results extend to SW-CRTs with non-continuous outcomes. Methods: In this article, we mathematically prove that the inclusion of unexposed clusters improves the precision of the fixed effects intervention effect estimator for SW-CRTs with binary and count outcomes. We then explore the benefits of including an unexposed cluster in simulated datasets with binary or count outcomes and a real palliative care data example with binary outcomes. Results: The simulations show that including unexposed clusters leads to tangible improvements in the precision, power, and root mean square error of the intervention effect estimator. The inclusion of the unexposed cluster in the SW-CRT of a novel palliative care intervention with binary outcomes yielded smaller standard errors and narrower 95% Wald Confidence Intervals. Conclusions: In this article, we demonstrate that the inclusion of unexposed clusters in the fixed effects analysis can lead to the improvements in precision, power, and RMSE of the fixed effects intervention effect estimator for SW-CRTs with binary or count outcomes.
- 28 October 2024
- Published by Springer Nature in BMC Medical Research Methodology
- Vol. 24 (1) , 1-11
Open AccessInclusion of unexposed clusters improves the precision of fixed effects analysis of stepped-wedge cluster randomized trials with binary and count outcomes
Background: The fixed effects model is a useful alternative to the mixed effects model for analyzing stepped-wedge cluster randomized trials (SW-CRTs). It controls for all time-invariant cluster-level confounders and has proper control of type I error when the number of clusters is small. While all clusters in a SW-CRT are typically designed to crossover from the control to receive the intervention, some trials can end with unexposed clusters (clusters that never receive the intervention), such as when a trial is terminated early due to safety concerns. It was previously unclear whether unexposed clusters would contribute to the estimation of the intervention effect in a fixed effects analysis. However, recent work has demonstrated that including an unexposed cluster can improve the precision of the intervention effect estimator in a fixed effects analysis of SW-CRTs with continuous outcomes. Still, SW-CRTs are commonly designed with binary outcomes and it is unknown if those previous results extend to SW-CRTs with non-continuous outcomes. Methods: In this article, we mathematically prove that the inclusion of unexposed clusters improves the precision of the fixed effects intervention effect estimator for SW-CRTs with binary and count outcomes. We then explore the benefits of including an unexposed cluster in simulated datasets with binary or count outcomes and a real palliative care data example with binary outcomes. Results: The simulations show that including unexposed clusters leads to tangible improvements in the precision, power, and root mean square error of the intervention effect estimator. The inclusion of the unexposed cluster in the SW-CRT of a novel palliative care intervention with binary outcomes yielded smaller standard errors and narrower 95% Wald Confidence Intervals. Conclusions: In this article, we demonstrate that the inclusion of unexposed clusters in the fixed effects analysis can lead to the improvements in precision, power, and RMSE of the fixed effects intervention effect estimator for SW-CRTs with binary or count outcomes.
- Open Access
- 25 October 2024
- Published by Springer Nature in Trials
- Vol. 25 (1) , 1-18
The fixed-effects model for robust analysis of stepped-wedge cluster trials with a small number of clusters and continuous outcomes: a simulation study
Background: Stepped-wedge cluster trials (SW-CTs) describe a cluster trial design where treatment rollout is staggered over the course of the trial. Clusters are commonly randomized to receive treatment beginning at different time points in this study design (commonly referred to as a Stepped-wedge cluster randomized trial; SW-CRT), but they can also be non-randomized. Trials with this design regularly have a low number of clusters and can be vulnerable to covariate imbalance. To address such covariate imbalance, previous work has examined covariate-constrained randomization and analysis adjustment for imbalanced covariates in mixed-effects models. These methods require the imbalanced covariate to always be known and measured. In contrast, the fixed-effects model automatically adjusts for all imbalanced time-invariant covariates, both measured and unmeasured, and has been implicated to have proper type I error control in SW-CTs with a small number of clusters and binary outcomes. Methods: We present a simulation study comparing the performance of the fixed-effects model against the mixed-effects model in randomized and non-randomized SW-CTs with small numbers of clusters and continuous outcomes. Additionally, we compare these models in scenarios with cluster-level covariate imbalances or confounding. Results: We found that the mixed-effects model can have low coverage probabilities and inflated type I error rates in SW-CTs with continuous outcomes, especially with a small number of clusters or when the ICC is low. Furthermore, mixed-effects models with a Satterthwaite or Kenward-Roger small sample correction can still result in inflated or overly conservative type I error rates, respectively. In contrast, the fixed-effects model consistently produced the target level of coverage probability and type I error rates without dramatically compromising power. Furthermore, the fixed-effects model was able to automatically account for all time-invariant cluster-level covariate imbalances and confounding to robustly yield unbiased estimates. Conclusions: We recommend the fixed-effects model for robust analysis of SW-CTs with a small number of clusters and continuous outcomes, due to its proper type I error control and ability to automatically adjust for all potential imbalanced time-invariant cluster-level covariates and confounders.
- 25 October 2024
- Published by Springer Nature in Trials
- Vol. 25 (1) , 1-18
Open AccessThe fixed-effects model for robust analysis of stepped-wedge cluster trials with a small number of clusters and continuous outcomes: a simulation study
Background: Stepped-wedge cluster trials (SW-CTs) describe a cluster trial design where treatment rollout is staggered over the course of the trial. Clusters are commonly randomized to receive treatment beginning at different time points in this study design (commonly referred to as a Stepped-wedge cluster randomized trial; SW-CRT), but they can also be non-randomized. Trials with this design regularly have a low number of clusters and can be vulnerable to covariate imbalance. To address such covariate imbalance, previous work has examined covariate-constrained randomization and analysis adjustment for imbalanced covariates in mixed-effects models. These methods require the imbalanced covariate to always be known and measured. In contrast, the fixed-effects model automatically adjusts for all imbalanced time-invariant covariates, both measured and unmeasured, and has been implicated to have proper type I error control in SW-CTs with a small number of clusters and binary outcomes. Methods: We present a simulation study comparing the performance of the fixed-effects model against the mixed-effects model in randomized and non-randomized SW-CTs with small numbers of clusters and continuous outcomes. Additionally, we compare these models in scenarios with cluster-level covariate imbalances or confounding. Results: We found that the mixed-effects model can have low coverage probabilities and inflated type I error rates in SW-CTs with continuous outcomes, especially with a small number of clusters or when the ICC is low. Furthermore, mixed-effects models with a Satterthwaite or Kenward-Roger small sample correction can still result in inflated or overly conservative type I error rates, respectively. In contrast, the fixed-effects model consistently produced the target level of coverage probability and type I error rates without dramatically compromising power. Furthermore, the fixed-effects model was able to automatically account for all time-invariant cluster-level covariate imbalances and confounding to robustly yield unbiased estimates. Conclusions: We recommend the fixed-effects model for robust analysis of SW-CTs with a small number of clusters and continuous outcomes, due to its proper type I error control and ability to automatically adjust for all potential imbalanced time-invariant cluster-level covariates and confounders.
- 4 March 2024
- Published by Wiley in Statistics in Medicine
- Vol. 43 (10) , 1955-1972
Estimation and reduction of bias in self‐controlled case series with non‐rare event dependent outcomes and heterogeneous populations
The self‐controlled case series (SCCS) is a commonly adopted study design in the assessment of vaccine and drug safety. Recurrent event data collected from SCCS studies are typically analyzed using the conditional Poisson model which assumes event times are independent within‐cases. This assumption is violated in the presence of event dependence, where the occurrence of an event influences the probability and timing of subsequent events. When event dependence is suspected in an SCCS study, the standard recommendation is to include only the first event from each case in the analysis. However, first event analysis can still yield biased estimates of the exposure relative incidence if the outcome event is not rare. We first demonstrate that the bias in first event analysis can be even higher than previously assumed when subpopulations with different baseline incidence rates are present and describe an improved method for estimating this bias. Subsequently, we propose a novel partitioned analysis method and demonstrate how it can reduce this bias. We provide a recommendation to guide the number of partitions to use with the partitioned analysis, illustrate this recommendation with an example SCCS study of the association between beta‐blockers and acute myocardial infarction, and compare the partitioned analysis against other SCCS analysis methods by simulation.
- 4 March 2024
- Published by Wiley in Statistics in Medicine
- Vol. 43 (10) , 1955-1972
Estimation and reduction of bias in self‐controlled case series with non‐rare event dependent outcomes and heterogeneous populations
The self‐controlled case series (SCCS) is a commonly adopted study design in the assessment of vaccine and drug safety. Recurrent event data collected from SCCS studies are typically analyzed using the conditional Poisson model which assumes event times are independent within‐cases. This assumption is violated in the presence of event dependence, where the occurrence of an event influences the probability and timing of subsequent events. When event dependence is suspected in an SCCS study, the standard recommendation is to include only the first event from each case in the analysis. However, first event analysis can still yield biased estimates of the exposure relative incidence if the outcome event is not rare. We first demonstrate that the bias in first event analysis can be even higher than previously assumed when subpopulations with different baseline incidence rates are present and describe an improved method for estimating this bias. Subsequently, we propose a novel partitioned analysis method and demonstrate how it can reduce this bias. We provide a recommendation to guide the number of partitions to use with the partitioned analysis, illustrate this recommendation with an example SCCS study of the association between beta‐blockers and acute myocardial infarction, and compare the partitioned analysis against other SCCS analysis methods by simulation.