Wim Van der Elst

Statistical evaluation of surrogate endpoints with examples from cancer clinical trials

A surrogate endpoint is intended to replace a clinical endpoint for the evaluation of new treatments when it can be measured more cheaply, more conveniently, more frequently, or earlier than that clinical endpoint. A surrogate endpoint is expected to predict clinical benefit, harm, or lack of these. Besides the biological plausibility of a surrogate, a quantitative assessment of the strength of evidence for surrogacy requires the demonstration of the prognostic value of the surrogate for the clinical outcome, and evidence that treatment effects on the surrogate reliably predict treatment effects on the clinical outcome. We focus on these two conditions, and outline the statistical approaches that have been proposed to assess the extent to which these conditions are fulfilled. When data are available from a single trial, one can assess the individual level association between the surrogate and the true endpoint. When data are available from several trials, one can additionally assess the trial level association between the treatment effect on the surrogate and the treatment effect on the true endpoint. In the latter case, the surrogate threshold effect can be estimated as the minimum effect on the surrogate endpoint that predicts a statistically significant effect on the clinical endpoint. All these concepts are discussed in the context of randomized clinical trials in oncology, and illustrated with two meta-analyses in gastric cancer.

Establishing normative data for multi-trial memory tests: the multivariate regression-based approach

Abstract Objective: Multi-trial memory tests are widely used in research and clinical practice because they allow for assessing different aspects of memory and learning in a single comprehensive test procedure. However, the use of multi-trial memory tests also raises some key data analysis issues. Indeed, the different trial scores are typically all correlated, and this correlation has to be properly accounted for in the statistical analyses. In the present paper, the focus is on the setting where normative data have to be established for multi-trial memory tests. At present, normative data for such tests are typically based on a series of univariate analyses, i.e. a statistical model is fitted for each of the test scores separately. This approach is suboptimal because (1) the correlated nature of the data is not accounted for, (2) multiple testing issues may arise, and (3) the analysis is not parsimonious. Method and results: Here, a normative approach that is not hampered by these issues is proposed (the so-called multivariate regression-based approach). The methodology is exemplified in a sample of N = 221 Dutch-speaking children (aged between 5.82 and 15.49 years) who were administered Rey’s Auditory Verbal Learning Test. An online Appendix that details how the analyses can be conducted in practice (using the R software) is also provided. Conclusion: The multivariate normative regression-based approach has some substantial methodological advantages over univariate regression-based methods. In addition, the method allows for testing substantive hypotheses that cannot be addressed in a univariate framework (e.g. trial by covariate interactions can be modeled).

Cognitive functioning in the general population: Factor structure and association with mental disorders—The neuropsychological test battery of the mental health module of the German Health Interview and Examination Survey for Adults (DEGS1‐MH)

The objective of this study is to obtain population level data about cognitive functions and their association with mental disorders. We here report factor analytic and psychometric findings of a neuropsychological test battery and examine the association of current and past mental disorders with cognitive function in a large nationwide population‐based sample of 18‐ to 79‐year‐old adults in Germany (n = 3,667) participating in the mental health module of the German Health Interview and Examination Survey for Adults 2008–2011. Confirmatory factor analysis confirmed verbal memory and executive function factors. Older age was strongly associated with lower verbal memory and executive function and with higher vocabulary scores. After adjustment for age, sex, and education, rather modest decrements were found for verbal memory (β = −.118, p = .002) and executive functions (β = −.191, p < .001) in participants with any current mental disorder (n = 442) compared to those without (n = 3,201). Small decrements in memory (β = −.064, p = .031) and executive function (β = −.111, p < .001) were found in participants with any mental disorder in the last 12 months but not in those with past (fully or partially remitted) mental disorders, compared to participants without a history of mental disorder. More fine‐grained analyses of these data will investigate the complex interplay between cognition, health behaviors, and specific mental and somatic diseases.

Predictive value of different covariates influencing pregnancy rate following intrauterine insemination with homologous semen: a prospective cohort study

The aim was to examine the value of different covariates in the prediction of intrauterine insemination (IUI) success. Between July 2011 and September 2015, data from 1401 IUI cycles with homologous semen in 556 couples were collected prospectively, by questionnaire, in a tertiary referral infertility centre. Statistical analysis was performed using generalized estimating equations (GEEs). GEEs were used instead of an ordinary logistic regression model to take into account the correlation between observations from the same person. The primary outcome parameter was clinical pregnancy rate (CPR), confirmed with a gestational sac and fetal heartbeat on ultrasonography at 7-8 weeks. An overall CPR of 9.5% per cycle was observed. Univariate statistical analysis revealed female and male age, male smoking, female body mass index, ovarian stimulation and inseminating motile count (IMC) as covariates significantly influencing CPR per cycle. Multivariate GEE analysis revealed that the only valuable prognostic covariates included female age, male smoking and infertility status (i.e. primary/secondary infertility). IMC showed a significant curvilinear relationship, with first an increase and then a decrease in pregnancy rate.

Predictive factors influencing pregnancy rates after intrauterine insemination with frozen donor semen: a prospective cohort study

The extent to which certain parameters can influence pregnancy rates after intrauterine insemination with frozen donor semen was examined prospectively. Between July 2011 and September 2015, 402 women received 1264 IUI cycles with frozen donor semen in a tertiary referral infertility centre. A case report form was used to collect data prospectively. The primary outcome measure was clinical pregnancy rate (CPR), confirmed by detection of a gestational sac and fetal heartbeat using ultrasonography at 7-8 weeks of gestation. Statistical analysis was carried out using generalized estimating equations (GEE) to account for the correlation between observations from the same patient. Overall, CPR per cycle was 17.2%. Multivariate GEE analysis revealed the following parameters as predictive for a successful pregnancy outcome: female age (P = 0.0003), non-smoking or smoking fewer than 15 cigarettes a day (P = 0.0470 and P = 0.0235, respectively), secondary infertility (P = 0.0062), low progesterone levels at day zero of the cycle (P = 0.0164) and use of ovarian stimulation with HMG and recombinant FSH compared with clomiphene citrate and natural cycle (P = 0.0006 and P = 0.0004, respectively). These parameters were the most important factors influencing the success rate in a sperm donation programme.

Compensating for memory losses throughout aging: Validation and normalization of the memory compensation questionnaire (MCQ) for non-clinical French populations

AIMnThe MCQ is a seven-factor scale that measures individual differences in the tendency to select particular strategies and to overcome perceived or real memory losses. Our aim was to establish a French version of the MCQ and to evaluate its psychometric properties in a lifespan perspective. We first tested the underlying factor structure of the MCQ in a large sample of 749 adults from aged from 18 to 92 years.nnnRESULTSnThe results showed that the factor structure of the French version corresponded well with the one obtained in English-, Dutch- and Spanish-speaking samples, supporting the cross-national robustness of the MCQ. We confirmed a seven-factor order model that supports the construct validity of the questionnaire. The reliabilities of the scales were good (α>.70) to acceptable (α=.66 and .62). Criterion validity was verified by means of significant correlations between health composites and MCQ subscales. Gender and Age affected most of the MCQ subscales but not the Level of Education (LE).nnnCONCLUSIONnThe MCQ revealed to be a heuristic tool for assessing daily compensatory behaviors that are developed in order to achieve successful aging. Thus, regression-based normative data and a user-friendly computer program were provided to facilitate scoring and norming by clinicians and researchers who need to assess daily compensatory behaviors.

A maximum entropy approach for the evaluation of surrogate endpoints based on causal inference: Maximum entropy approach and surrogate endpoints

The maximum entropy principle offers a constructive criterion for setting up probability distributions on the basis of partial knowledge. In the present work, the principle is applied to tackle an important problem in the surrogate marker field, namely, the evaluation of a binary outcome as a putative surrogate for a binary true endpoint within a causal inference framework. In the first step, the maximum entropy principle is used to determine the relative frequencies associated with the values of the vector of potential outcomes. Subsequently, in the second step, these relative frequencies are used in combination with two newly proposed metrics of surrogacy, the so-called individual causal association and the surrogate predictive function, to assess the validity of the surrogate. The procedure is conceptually similar to the use of noninformative or reference priors in Bayesian statistics. Additionally, approximate, identifiable bounds are proposed for the estimands of interest, and their performance is studied via simulations. The methods are illustrated using data from a clinical trial involving schizophrenic patients, and a newly developed and user-friendly R package Surrogate is provided to carry out the validation exercise.

Beyond global differences between monolingual and bilingual children on the nonword repetition task: retention skills for phonemes’ identity and serial order

This study compared NRT-performance in monolingual Dutch and bilingual Turkish–Dutch third-graders using a Dutch Nonword Repetition Task (NRT). Several novel response analyses at the phoneme level were applied to further understand the earlier reported overall accuracy differences in NRT-performance between bilinguals and monolinguals. Analyses in which the retention of phonemes and the retention of their serial order were disentangled revealed that monolinguals outperform bilinguals with respect to the retention of the phonemes themselves. However, both groups did not differ in their retention of the serial order of correctly recalled phonemes. Furthermore, this study confirms that expressive vocabulary skills do affect overall NRT-performance. The results are discussed in light of current short-term memory (STM) models and the role of long-term phonological knowledge in NRT tasks.