Gerard J. P. van Breukelen
Current Organizations / Institutes
The organizations/institutes from the scholar's latest article
Previous Organizations / Institutes
The organizations / institutes from the scholar's other articles in the past ten years
Gerard J. P. van Breukelen has a total of 452 co-authors. Here are the top co-authors by number of shared publications:
- Math J. J. M. Candel
Department of Methodology and Statistics, Care and Public Health Research Institute (CAPHRI), Maastricht University, Maastricht, The Netherlands
17 shared publications Gerjo Kok
9 shared publications- Johan W.S. Vlaeyen
Experimental Health Psychology, Maastricht University, Maastricht, The Netherlands
9 shared publications Jelle Jolles
Maastricht University , The Netherlands
8 shared publicationsMartijn P. F. Berger
Department of Methodology and Statistics, Maastricht University, Maastricht, the Netherlands
7 shared publicationsMartijn P.F. Berger
6 shared publications
Latest Publications
- Open Access
- 21 July 2026
- Published by Wiley in Statistics in Medicine
- Vol. 45 (18-19) , e70677
Designing Efficient Multicenter Trials With a Maximin Strategy
For two‐treatment multicenter trials with a continuous outcome, designs are presented that minimize the number of subjects required or the research budget, when aiming for a desired power level. These designs optimize both the allocation ratio of participants between treatment and control condition within each center, and the balance between the number of centers and the number of participants per center. Because optimal designs rely on prior knowledge of variance parameters from the analysis model, which are unknown during the design stage, maximin designs are introduced. These designs guarantee a pre‐specified power level across plausible ranges of the unknown parameters and maximize power under worst‐case parameter values. The present study extends the existing literature on multicenter designs by deriving maximin designs not only for the case where the number of centers and the sample size per center can be optimized, but also when either the number of centers or the sample size per center is fixed due to practical constraints. An empirical example illustrates sample size calculation for such designs and potential budget savings. Additionally, an interactive R Shiny app has been developed to simplify sample size calculations for maximin multicenter trials.
- 21 July 2026
- Published by Wiley in Statistics in Medicine
- Vol. 45 (18-19) , e70677
Open AccessDesigning Efficient Multicenter Trials With a Maximin Strategy
For two‐treatment multicenter trials with a continuous outcome, designs are presented that minimize the number of subjects required or the research budget, when aiming for a desired power level. These designs optimize both the allocation ratio of participants between treatment and control condition within each center, and the balance between the number of centers and the number of participants per center. Because optimal designs rely on prior knowledge of variance parameters from the analysis model, which are unknown during the design stage, maximin designs are introduced. These designs guarantee a pre‐specified power level across plausible ranges of the unknown parameters and maximize power under worst‐case parameter values. The present study extends the existing literature on multicenter designs by deriving maximin designs not only for the case where the number of centers and the sample size per center can be optimized, but also when either the number of centers or the sample size per center is fixed due to practical constraints. An empirical example illustrates sample size calculation for such designs and potential budget savings. Additionally, an interactive R Shiny app has been developed to simplify sample size calculations for maximin multicenter trials.
- Open Access
- 16 April 2026
- Published by Wiley in Statistics in Medicine
- Vol. 45 (8-9) , e70549
Correcting for a Baseline Difference in Group Comparisons on a Binary Outcome: Covariate Adjustment or Analysis of Change?
Effects of treatments or exposures are evaluated by comparing a treated or exposed group with a control group after treatment or exposure. A baseline group difference can be accounted for by covariate adjustment or by analyzing change from baseline. For quantitative outcomes these two methods can give contradictory results (Lord's paradox) and especially covariate adjustment is questionable for nonrandomized group comparisons. This paper explores analogous methods for the case of a binary outcome, specifically logistic regression of the outcome on group and baseline, ordinal regression of change from baseline on group, and mixed logistic regression and generalized estimating equations ( GEE ) for repeated measures. The methods were compared mathematically, conceptually in terms of causal diagrams and estimands, and numerically on fictitious scenarios that varied in whether groups differed at baseline and/or in change over time on the logodds scale. The methods were also compared on a smoking prevention study among school children. The scenario results were similar to those published for quantitative outcomes: Logistic regression of outcome on group and baseline gave almost the same results as mixed logistic regression and GEE without parameter for a baseline group difference. Ordinal regression of change from baseline gave almost the same results as mixed logistic regression and GEE allowing for a baseline group difference. These (near‐)equivalences are in line with two mathematical proofs in this paper. Further, in data sets with a baseline group difference, covariate adjustment and change analysis led to contradictory conclusions. The results from the smoking prevention study confirmed the above results and showed Lord's paradox.
- 16 April 2026
- Published by Wiley in Statistics in Medicine
- Vol. 45 (8-9) , e70549
Open AccessCorrecting for a Baseline Difference in Group Comparisons on a Binary Outcome: Covariate Adjustment or Analysis of Change?
Effects of treatments or exposures are evaluated by comparing a treated or exposed group with a control group after treatment or exposure. A baseline group difference can be accounted for by covariate adjustment or by analyzing change from baseline. For quantitative outcomes these two methods can give contradictory results (Lord's paradox) and especially covariate adjustment is questionable for nonrandomized group comparisons. This paper explores analogous methods for the case of a binary outcome, specifically logistic regression of the outcome on group and baseline, ordinal regression of change from baseline on group, and mixed logistic regression and generalized estimating equations ( GEE ) for repeated measures. The methods were compared mathematically, conceptually in terms of causal diagrams and estimands, and numerically on fictitious scenarios that varied in whether groups differed at baseline and/or in change over time on the logodds scale. The methods were also compared on a smoking prevention study among school children. The scenario results were similar to those published for quantitative outcomes: Logistic regression of outcome on group and baseline gave almost the same results as mixed logistic regression and GEE without parameter for a baseline group difference. Ordinal regression of change from baseline gave almost the same results as mixed logistic regression and GEE allowing for a baseline group difference. These (near‐)equivalences are in line with two mathematical proofs in this paper. Further, in data sets with a baseline group difference, covariate adjustment and change analysis led to contradictory conclusions. The results from the smoking prevention study confirmed the above results and showed Lord's paradox.
- Open Access
- 13 March 2026
- Published by SAGE Publications in Statistical Methods in Medical Research
- Vol. 35 (4) , 695-712
Efficient design of partially nested randomized trials: A maximin approach
For two-treatment randomized trials with clustering in one of the treatment arms and a continuous outcome, designs are presented that minimize the number of subjects or the amount of research budget, when aiming for a desired power level. These designs optimize the treatment-to-control allocation ratio of study participants but also optimize the choice between the number of clusters (such as therapy groups) versus the number of persons per cluster (therapy group) in the arm with clustering. Optimal designs require prior knowledge of parameters from the analysis model, which are unknown during the design stage. We present maximin designs which address this by ensuring a pre-specified power level for plausible ranges of the unknown parameters, while maximizing the power for worst-case values of these parameters. Maximin designs are also derived when the number of clusters, or the cluster size is fixed due to practical constraints. An empirical example illustrates how to calculate sample sizes for such practical designs and shows how much these maximin designs can reduce the required research budgets compared to designs with equal subject numbers in treatment and control. A user-friendly R Shiny app facilitates these sample size calculations.
- 13 March 2026
- Published by SAGE Publications in Statistical Methods in Medical Research
- Vol. 35 (4) , 695-712
Open AccessEfficient design of partially nested randomized trials: A maximin approach
For two-treatment randomized trials with clustering in one of the treatment arms and a continuous outcome, designs are presented that minimize the number of subjects or the amount of research budget, when aiming for a desired power level. These designs optimize the treatment-to-control allocation ratio of study participants but also optimize the choice between the number of clusters (such as therapy groups) versus the number of persons per cluster (therapy group) in the arm with clustering. Optimal designs require prior knowledge of parameters from the analysis model, which are unknown during the design stage. We present maximin designs which address this by ensuring a pre-specified power level for plausible ranges of the unknown parameters, while maximizing the power for worst-case values of these parameters. Maximin designs are also derived when the number of clusters, or the cluster size is fixed due to practical constraints. An empirical example illustrates how to calculate sample sizes for such practical designs and shows how much these maximin designs can reduce the required research budgets compared to designs with equal subject numbers in treatment and control. A user-friendly R Shiny app facilitates these sample size calculations.
- 15 October 2025
- Published by BMJ in Journal of Medical Ethics
Keep calm and think twice: methodological issues with the recommendations by Giordano (2025)
- 15 October 2025
- Published by BMJ in Journal of Medical Ethics
Keep calm and think twice: methodological issues with the recommendations by Giordano (2025)
- Open Access
- 4 April 2025
- Published by Taylor & Francis in European Journal of Developmental Psychology
- p. 1-21
How to improve research methodology in gender care: a non-binary choice
- 4 April 2025
- Published by Taylor & Francis in European Journal of Developmental Psychology
- p. 1-21
Open AccessHow to improve research methodology in gender care: a non-binary choice
- 13 February 2025
- Published by American Psychological Association (APA) in Psychological Methods
Efficient design of cluster randomized trials and individually randomized group treatment trials.
For cluster randomized trials and individually randomized group treatment trials that compare two treatments on a continuous outcome, designs are presented that minimize the number of subjects or the amount of research budget, when aiming for a desired power level. These designs optimize the treatment-to-control allocation ratio of study participants but also optimize the choice between the number of clusters/groups versus the number of persons per cluster/group. Given that optimal designs require prior knowledge of parameters from the analysis model, which are often unknown during the design stage-especially outcome variances-maximin designs are introduced. These designs ensure a prespecified power level for plausible ranges of the unknown parameters and maximize power for the worst-case values of these parameters. The present study not only reviews but also extends the existing literature by deriving optimal and maximin designs when the number of clusters/groups are fixed because of practical constraints. How to calculate sample sizes in such practical designs and how much budget may be saved are illustrated for an empirical example. To facilitate sample size calculation for each of the variants of the maximin designs considered, an easy-to-use interactive R Shiny app has been developed and made available. (PsycInfo Database Record (c) 2025 APA, all rights reserved).
- 13 February 2025
- Published by American Psychological Association (APA) in Psychological Methods
Efficient design of cluster randomized trials and individually randomized group treatment trials.
For cluster randomized trials and individually randomized group treatment trials that compare two treatments on a continuous outcome, designs are presented that minimize the number of subjects or the amount of research budget, when aiming for a desired power level. These designs optimize the treatment-to-control allocation ratio of study participants but also optimize the choice between the number of clusters/groups versus the number of persons per cluster/group. Given that optimal designs require prior knowledge of parameters from the analysis model, which are often unknown during the design stage-especially outcome variances-maximin designs are introduced. These designs ensure a prespecified power level for plausible ranges of the unknown parameters and maximize power for the worst-case values of these parameters. The present study not only reviews but also extends the existing literature by deriving optimal and maximin designs when the number of clusters/groups are fixed because of practical constraints. How to calculate sample sizes in such practical designs and how much budget may be saved are illustrated for an empirical example. To facilitate sample size calculation for each of the variants of the maximin designs considered, an easy-to-use interactive R Shiny app has been developed and made available. (PsycInfo Database Record (c) 2025 APA, all rights reserved).
- Open Access
- 27 May 2024
- Published by Taylor & Francis in Journal of Clinical and Experimental Neuropsychology
- Vol. 46 (5) , 449-465
Young children’s performance on a design fluency task: longitudinal data on total number of designs, clustering and switching, and regression-based norms
IntroductionDesign fluency (DF) tasks are commonly used to assess executive functions such as attentional control, cognitive flexibility, self-monitoring and strategy use. Next to the total number of correct designs, the standard outcome of a DF task, clustering and switching can help disentangle the processes underlying DF performance. We present the first longitudinal study of 4-8-year-old children's developmental DF trajectories.MethodAt initial enrollment, children (n = 228) were aged between 4.05 and 6.88 years (M = 5.18, SD = 0.59) and attended Dutch primary schools. The DF task was administered at three time points, each time point separated by approximately 1 year. Data were analyzed using mixed regression for total number of correct designs and switching, and mixed logistic regression analysis for clustering.ResultsThe total number of correct designs increased linearly across the three time points. Across all time points, children made very few clusters, and most clusters consisted of only 3 designs. Clustering only increased at the third assessment compared to the two previous assessments. Switching increased up to the second assessment, but not after that. The number of switches was highly correlated with the total number of correct designs at all time points (r = 0.78 to r = 0.85). These developmental trajectories were similar for all children regardless of their baseline age. Normative data are given for the total number of correct designs and switching.ConclusionsChildren as of age 4 onwards can perform a DF task. For children as young as 4-8 years old, computing clustering, and switching measures is of limited value to study cognitive processes underlying DF performance, next to the total number of correct designs. There were no sex differences on any of the DF outcomes. Level of parental education (LPE) was positively associated with the total number of correct designs and switching.
- 27 May 2024
- Published by Taylor & Francis in Journal of Clinical and Experimental Neuropsychology
- Vol. 46 (5) , 449-465
Open AccessYoung children’s performance on a design fluency task: longitudinal data on total number of designs, clustering and switching, and regression-based norms
IntroductionDesign fluency (DF) tasks are commonly used to assess executive functions such as attentional control, cognitive flexibility, self-monitoring and strategy use. Next to the total number of correct designs, the standard outcome of a DF task, clustering and switching can help disentangle the processes underlying DF performance. We present the first longitudinal study of 4-8-year-old children's developmental DF trajectories.MethodAt initial enrollment, children (n = 228) were aged between 4.05 and 6.88 years (M = 5.18, SD = 0.59) and attended Dutch primary schools. The DF task was administered at three time points, each time point separated by approximately 1 year. Data were analyzed using mixed regression for total number of correct designs and switching, and mixed logistic regression analysis for clustering.ResultsThe total number of correct designs increased linearly across the three time points. Across all time points, children made very few clusters, and most clusters consisted of only 3 designs. Clustering only increased at the third assessment compared to the two previous assessments. Switching increased up to the second assessment, but not after that. The number of switches was highly correlated with the total number of correct designs at all time points (r = 0.78 to r = 0.85). These developmental trajectories were similar for all children regardless of their baseline age. Normative data are given for the total number of correct designs and switching.ConclusionsChildren as of age 4 onwards can perform a DF task. For children as young as 4-8 years old, computing clustering, and switching measures is of limited value to study cognitive processes underlying DF performance, next to the total number of correct designs. There were no sex differences on any of the DF outcomes. Level of parental education (LPE) was positively associated with the total number of correct designs and switching.
- Top Cited PapersOpen Access
- 20 May 2024
- Published by American Society of Clinical Oncology (ASCO) in Journal of Clinical Oncology
- Vol. 42 (15) , 1799-1809
Laparoscopic Versus Open Hemihepatectomy: The ORANGE II PLUS Multicenter Randomized Controlled Trial
PURPOSE: To compare outcomes after laparoscopic versus open major liver resection (hemihepatectomy) mainly for primary or metastatic cancer. The primary outcome measure was time to functional recovery. Secondary outcomes included morbidity, quality of life (QoL), and for those with cancer, resection margin status and time to adjuvant systemic therapy. PATIENTS AND METHODS: This was a multicenter, randomized controlled, patient-blinded, superiority trial on adult patients undergoing hemihepatectomy. Patients were recruited from 16 hospitals in Europe between November 2013 and December 2018. RESULTS: Of the 352 randomly assigned patients, 332 patients (94.3%) underwent surgery (laparoscopic, n = 166 and open, n = 166) and comprised the analysis population. The median time to functional recovery was 4 days (IQR, 3-5; range, 1-30) for laparoscopic hemihepatectomy versus 5 days (IQR, 4-6; range, 1-33) for open hemihepatectomy (difference, –17.5% [96% CI, –25.6 to –8.4]; P < .001). There was no difference in major complications (laparoscopic 24/166 [14.5%] v open 28/166 [16.9%]; odds ratio [OR], 0.84; P = .58). Regarding QoL, both global health status (difference, 3.2 points; P < .001) and body image (difference, 0.9 points; P < .001) scored significantly higher in the laparoscopic group. For the 281 (84.6%) patients with cancer, R0 resection margin status was similar (laparoscopic 106 [77.9%] v open 122 patients [84.1%], OR, 0.60; P = .14) with a shorter time to adjuvant systemic therapy in the laparoscopic group (46.5 days v 62.8 days, hazard ratio, 2.20; P = .009). CONCLUSION: Among patients undergoing hemihepatectomy, the laparoscopic approach resulted in a shorter time to functional recovery compared with open surgery. In addition, it was associated with a better QoL, and in patients with cancer, a shorter time to adjuvant systemic therapy with no adverse impact on cancer outcomes observed.
- 20 May 2024
- Published by American Society of Clinical Oncology (ASCO) in Journal of Clinical Oncology
- Vol. 42 (15) , 1799-1809
Top Cited PapersOpen AccessLaparoscopic Versus Open Hemihepatectomy: The ORANGE II PLUS Multicenter Randomized Controlled Trial
PURPOSE: To compare outcomes after laparoscopic versus open major liver resection (hemihepatectomy) mainly for primary or metastatic cancer. The primary outcome measure was time to functional recovery. Secondary outcomes included morbidity, quality of life (QoL), and for those with cancer, resection margin status and time to adjuvant systemic therapy. PATIENTS AND METHODS: This was a multicenter, randomized controlled, patient-blinded, superiority trial on adult patients undergoing hemihepatectomy. Patients were recruited from 16 hospitals in Europe between November 2013 and December 2018. RESULTS: Of the 352 randomly assigned patients, 332 patients (94.3%) underwent surgery (laparoscopic, n = 166 and open, n = 166) and comprised the analysis population. The median time to functional recovery was 4 days (IQR, 3-5; range, 1-30) for laparoscopic hemihepatectomy versus 5 days (IQR, 4-6; range, 1-33) for open hemihepatectomy (difference, –17.5% [96% CI, –25.6 to –8.4]; P < .001). There was no difference in major complications (laparoscopic 24/166 [14.5%] v open 28/166 [16.9%]; odds ratio [OR], 0.84; P = .58). Regarding QoL, both global health status (difference, 3.2 points; P < .001) and body image (difference, 0.9 points; P < .001) scored significantly higher in the laparoscopic group. For the 281 (84.6%) patients with cancer, R0 resection margin status was similar (laparoscopic 106 [77.9%] v open 122 patients [84.1%], OR, 0.60; P = .14) with a shorter time to adjuvant systemic therapy in the laparoscopic group (46.5 days v 62.8 days, hazard ratio, 2.20; P = .009). CONCLUSION: Among patients undergoing hemihepatectomy, the laparoscopic approach resulted in a shorter time to functional recovery compared with open surgery. In addition, it was associated with a better QoL, and in patients with cancer, a shorter time to adjuvant systemic therapy with no adverse impact on cancer outcomes observed.
- Open Access
- 17 August 2023
- Published by Taylor & Francis in Multivariate Behavioral Research
- Vol. 59 (2) , 206-228
Cluster Randomized Trials with a Pretest and Posttest: Equivalence of Three-, Two- and One-Level Analyses, and Sample Size Calculation
In a cluster randomized trial clusters of persons, for instance, schools or health centers, are assigned to treatments, and all persons in the same cluster get the same treatment. Although less powerful than individual randomization, cluster randomization is a good alternative if individual randomization is impossible or leads to severe treatment contamination (carry-over). Focusing on cluster randomized trials with a pretest and post-test of a quantitative outcome, this paper shows the equivalence of four methods of analysis: a three-level mixed (multilevel) regression for repeated measures with as levels cluster, person, and time, and allowing for unstructured between-cluster and within-cluster covariance matrices; a two-level mixed regression with as levels cluster and person, using change from baseline as outcome; a two-level mixed regression with as levels cluster and time, using cluster means as data; a one-level analysis of cluster means of change from baseline. Subsequently, similar equivalences are shown between a constrained mixed model and methods using the pretest as covariate. All methods are also compared on a cluster randomized trial on mental health in children. From these equivalences follows a simple method to calculate the sample size for a cluster randomized trial with baseline measurement, which is demonstrated step-by-step.
- 17 August 2023
- Published by Taylor & Francis in Multivariate Behavioral Research
- Vol. 59 (2) , 206-228
Open AccessCluster Randomized Trials with a Pretest and Posttest: Equivalence of Three-, Two- and One-Level Analyses, and Sample Size Calculation
In a cluster randomized trial clusters of persons, for instance, schools or health centers, are assigned to treatments, and all persons in the same cluster get the same treatment. Although less powerful than individual randomization, cluster randomization is a good alternative if individual randomization is impossible or leads to severe treatment contamination (carry-over). Focusing on cluster randomized trials with a pretest and post-test of a quantitative outcome, this paper shows the equivalence of four methods of analysis: a three-level mixed (multilevel) regression for repeated measures with as levels cluster, person, and time, and allowing for unstructured between-cluster and within-cluster covariance matrices; a two-level mixed regression with as levels cluster and person, using change from baseline as outcome; a two-level mixed regression with as levels cluster and time, using cluster means as data; a one-level analysis of cluster means of change from baseline. Subsequently, similar equivalences are shown between a constrained mixed model and methods using the pretest as covariate. All methods are also compared on a cluster randomized trial on mental health in children. From these equivalences follows a simple method to calculate the sample size for a cluster randomized trial with baseline measurement, which is demonstrated step-by-step.
- 1 February 2023
- Published by American Psychological Association (APA) in Psychological Methods
- Vol. 28 (1) , 89-106
Sample size calculation and optimal design for regression-based norming of tests and questionnaires.
To prevent mistakes in psychological assessment, the precision of test norms is important. This can be achieved by drawing a large normative sample and using regression-based norming. Based on that norming method, a procedure for sample size planning to make inference on Z-scores and percentile rank scores is proposed. Sampling variance formulas for these norm statistics are derived and used to obtain the optimal design, that is, the optimal predictor distribution, for the normative sample, thereby maximizing precision of estimation. This is done under five regression models with a quantitative and a categorical predictor, differing in whether they allow for interaction and nonlinearity. Efficient robust designs are given in case of uncertainty about the regression model. Furthermore, formulas are provided to compute the normative sample size such that individuals' positions relative to the derived norms can be assessed with prespecified power and precision. (PsycInfo Database Record (c) 2023 APA, all rights reserved).
- 1 February 2023
- Published by American Psychological Association (APA) in Psychological Methods
- Vol. 28 (1) , 89-106
Sample size calculation and optimal design for regression-based norming of tests and questionnaires.
To prevent mistakes in psychological assessment, the precision of test norms is important. This can be achieved by drawing a large normative sample and using regression-based norming. Based on that norming method, a procedure for sample size planning to make inference on Z-scores and percentile rank scores is proposed. Sampling variance formulas for these norm statistics are derived and used to obtain the optimal design, that is, the optimal predictor distribution, for the normative sample, thereby maximizing precision of estimation. This is done under five regression models with a quantitative and a categorical predictor, differing in whether they allow for interaction and nonlinearity. Efficient robust designs are given in case of uncertainty about the regression model. Furthermore, formulas are provided to compute the normative sample size such that individuals' positions relative to the derived norms can be assessed with prespecified power and precision. (PsycInfo Database Record (c) 2023 APA, all rights reserved).