Biomedical subjects

Geert Verbeke

Publications and source records attributed to Geert Verbeke.

14 recordsLinked to original sources

PubMed2006

Is anaemia a risk factor for delirium in an acute geriatric population?

BACKGROUND/OBJECTIVE: Delirium is a common clinical problem in elderly patients. We aim to investigate whether anaemia is a risk factor for delirium in a hospitalized geriatric population. METHODS: During a 5-month prospective study, we investigated 190 elderly patients aged 70 years and older with a baseline Mini-Mental State Examination (short version), Confusion Assessment Method, demographic, clinical and laboratory data. RESULTS: Thirty-four patients were identified as delirious and 95 as anaemic according to the WHO criteria. Stepwise logistic regression revealed that anaemia (haemoglobin level <12 g/l in women and <13 g/l in men), male sex and a diagnosis of dementia were independent risk factors for delirium in the total study group. After adjustment for sex, age, diagnosis of dementia and dehydration, the odds ratio (OR) for anaemia (2.4; 95 CI = 1.02-5.54) remained significantly associated with delirium. When the study population was classified in groups according to sex, anaemia remained a significant risk factor for delirium in men (OR = 3.7; 95% CI = 1.03-15.6) after adjustment for the multiple variables but not in women (OR = 1.54; 95% CI = 0.48-4.9). When the haemoglobin levels were stratified into sex-specific quartiles, the adjusted OR for delirium for men with a haemoglobin level less than 11.1 g/dl was 13.1 (95% CI = 1.17-146). CONCLUSION: Anaemia is an independent risk factor for delirium and adds valuable information to previously validated predictive models in men but not in women and lower haemoglobin levels were associated with higher risk levels.

Aged↗

PubMed2006

Pairwise fitting of mixed models for the joint modeling of multivariate longitudinal profiles.

A mixed model is a flexible tool for joint modeling purposes, especially when the gathered data are unbalanced. However, computational problems due to the dimension of the joint covariance matrix of the random effects arise as soon as the number of outcomes and/or the number of used random effects per outcome increases. We propose a pairwise approach in which all possible bivariate models are fitted, and where inference follows from pseudo-likelihood arguments. The approach is applicable for linear, generalized linear, and nonlinear mixed models, or for combinations of these. The methodology will be illustrated for linear mixed models in the analysis of 22-dimensional, highly unbalanced, longitudinal profiles of hearing thresholds.

Auditory Threshold↗

PubMed2006

Statistical inference in generalized linear mixed models: a review.

We present a review of statistical inference in generalized linear mixed models (GLMMs). GLMMs are an extension of generalized linear models and are suitable for the analysis of non-normal data with a clustered structure. A GLMM contains parameters common to all clusters (fixed regression effects and variance components) and cluster-specific parameters. The latter parameters are assumed to be randomly drawn from a population distribution. The parameters of this population distribution (the variance components) have to be estimated together with the fixed effects. We focus on the case in which the cluster-specific parameters are normally distributed. The cluster-specific effects are integrated out of the likelihood so that the fixed effects and variance components can be estimated. Unfortunately, the integral over the cluster-specific effects is intractable for most GLMMs with a normal mixing distribution. Within a classical statistical framework, we distinguish between two broad classes of methods to handle this intractable integral: methods that rely on a numerical approximation to the integral and methods that use an analytical approximation to the integrand. Finally, we present an overview of available methods for testing hypotheses about the parameters of GLMMs.

Analysis of Variance↗

PubMed2005

On the relationship between all-cause, cardiovascular, cancer and residual mortality rates with age.

BACKGROUND: The existence of a highly significant linear relationship between the natural logarithm (ln) all-cause mortality rate and age at the population level is firmly established (r>0.99). The slope and intercept of the equation, however, vary markedly between populations. Whether this relationship also applies to specific disease entities has not been established. METHODS: Use was made of mortality rates for all-cause, total cardiovascular, total cancer and residual diseases. The midpoint of 5-year age classes between the ages of 35 and 84 years, obtained for both sexes, were analysed. The mean of the three latest available years, from the period 1997-1999 were used. RESULTS: The relationship also applies to a slightly lesser degree to the relationship between total cardiovascular mortality rate, consisting predominantly of ischemic heart disease and stroke, and age (r>0.99). Marginally better relationships are obtained using a second-degree polynomial equation between ln all-cause mortality rate and age, age as independent variables. Total ln cancer mortality rate, however, behaves differently with a significant negative deviation of the mortality rate from linearity at older ages. Residual mortality (non-cancer, non-cardiovascular) mortality shows a mirror pattern to cancer mortality. This residual mortality expressed as a percentage of all-cause mortality varies markedly between populations. The level of some major constituents of the residual mortality rates (respiratory diseases, pneumonia, ill-defined causes and senility) also varies markedly. CONCLUSIONS: The magnitude of the variation suggests misclassification or misdiagnosis of several important disease entities, for example, between senility and stroke or between pneumonia and lung cancer. This questions the validity of disease-specific mortality rates especially at older ages, making their comparison between countries less reliable.

Adult↗

PubMed2005

Multiple imputation for model checking: completed-data plots with missing and latent data.

In problems with missing or latent data, a standard approach is to first impute the unobserved data, then perform all statistical analyses on the completed dataset--corresponding to the observed data and imputed unobserved data--using standard procedures for complete-data inference. Here, we extend this approach to model checking by demonstrating the advantages of the use of completed-data model diagnostics on imputed completed datasets. The approach is set in the theoretical framework of Bayesian posterior predictive checks (but, as with missing-data imputation, our methods of missing-data model checking can also be interpreted as "predictive inference" in a non-Bayesian context). We consider the graphical diagnostics within this framework. Advantages of the completed-data approach include: (1) One can often check model fit in terms of quantities that are of key substantive interest in a natural way, which is not always possible using observed data alone. (2) In problems with missing data, checks may be devised that do not require to model the missingness or inclusion mechanism; the latter is useful for the analysis of ignorable but unknown data collection mechanisms, such as are often assumed in the analysis of sample surveys and observational studies. (3) In many problems with latent data, it is possible to check qualitative features of the model (for example, independence of two variables) that can be naturally formalized with the help of the latent data. We illustrate with several applied examples.

Animals↗

PubMed2004

Joint modelling of multivariate longitudinal profiles: pitfalls of the random-effects approach.

Due to its flexibility, the random-effects approach for the joint modelling of multivariate longitudinal profiles received a lot of attention in recent publications. In this approach different mixed models are joined by specifying a common distribution for their random-effects. Parameter estimates of this common distribution can then be used to evaluate the relation between the different responses. Using bivariate longitudinal measurements on pure-tone hearing thresholds, it will be shown that such a random-effects approach can yield misleading results for evaluating this relationship.

Adolescent↗

PubMed2004

Early and repetitive stimulation of the arm can substantially improve the long-term outcome after stroke: a 5-year follow-up study of a randomized trial.

BACKGROUND AND PURPOSE: Several studies have investigated the effect of therapeutic interventions for the arm in the acute phase after stroke, with follow-ups at a maximum of 12 months. The aim of this study was to examine the effect of repetitive sensorimotor training of the arm at 5 years after stroke. SUBJECTS AND METHODS: One hundred consecutive stroke patients were randomly allocated either to an experimental group that received daily additional sensorimotor stimulation of the arm or to a control group. The intervention period was 6 weeks. Assessments of the patients were made before, midway, and after intervention, and at 6 and 12 months after stroke. In this study, 62 patients were reassessed at 5 years after stroke. The Brunnström-Fugl-Meyer (BFM) test, Action Research Arm (ARA) test, and Barthel index (BI) were used as the primary outcome measures. RESULTS: At the 5-year follow-up, there was a statistically significant difference for both the BFM and ARA tests in favor of the experimental group. The mean differences in improvement between the groups from the initial evaluation to the 5-year assessment corresponded to 17 points on the BFM and 17.4 on the ARA. No effect was found for the BI. The treatment was most effective in patients with a severe initial motor deficit. CONCLUSIONS: Adding a specific intervention for the arm during the acute phase after a stroke resulted in a clinically meaningful and long-lasting effect on motor function. The effect can be attributed to early, repetitive, and targeted stimulation.

Adult↗

PubMed2004

Sensitivity analysis for pattern mixture models.

Incomplete series of data is a common feature in quality-of-life studies, in particular in chronic diseases where attrition of patients is high. Two alternative approaches to modeling longitudinal data with incomplete measurements have frequently been proposed in the literature, selection models and pattern-mixture models. In this paper we focus on, by way of sensitivity analysis, extrapolating incomplete patterns using identifying restrictions. Perhaps the best known ones are so-called complete case missing value restrictions (CCMV), where for a given pattern, the conditional distribution of the missing data, given the observed data, is equated to its counterpart in the completers. Available case missing value (ACMV) restrictions equate this conditional density to the one calculated from the subgroup of all patterns for which all required components have been observed. Neighboring case missing value restrictions (NCMV) equate this conditional density to the one calculated from the the pattern with one additional measurement obtained. In this paper, these three identifying restriction strategies are used to multiply impute missing data in a study in metastatic prostate cancer. Multiple imputation is employed to reduce the uncertainty of single imputation. It is shown how hypothesis testing and sensitivity analyses are carried out in this setting.

Humans↗

PubMed2004

Prospective study on late consequences of subclinical non-compliance with immunosuppressive therapy in renal transplant patients.

In this prospective study we compared the incidence of late acute rejections (LAR) and changes in serum-creatinine over time between compliers and noncompliers with immunosuppressive therapy more than 1 year post transplantation and explored the relative contribution of non-compliance and other risk factors in the occurrence of LAR. One hundred and forty-six adult renal transplant recipients were followed during a 5-year period. Patients were interviewed at the beginning of the study and categorized as non-compliers if they admitted to have skipped immunosuppressive medication on a regular basis during the previous 12 months. The occurrence of LAR during the follow-up period was recorded. We identified 22.6% non-compliers of which 21.2% experienced a late acute rejection compared with 8% in the group of compliers at 5 years postinclusion (p < 0.05). Kaplan-Meier survival analysis showed a decreased rejection free time in non-compliers compared with compliers (p = 0.03). Non-compliant patients had a 3.2 higher risk of LAR (Cox regression analysis, p = 0.005). Non-compliers experienced a higher increase in serum-creatinine over time (Linear Mixed Models, p < 0.001). Non-compliance in renal transplant patients more than 1-year post transplantation is associated with an increased risk for LAR and a higher increase in serum-creatinine during the following 5 years.

Adult↗

PubMed2003

A comparison of group sequential methods for binary longitudinal data.

Interim analyses are conducted to allow for early termination of the trial, for ethical as well as economical reasons. Here we consider interim analyses in repeated measurements studies where the measurements are binary. Two methods for analysing this kind of data are compared according to their operating characteristics. A subject-specific approach based on the logistic random-effects model is compared with the population-averaged approach based on the generalized estimating equations. The comparison is illustrated with simulations using a randomized clinical trial for toenail fungal infection.

Antifungal Agents↗

PubMed2003

The use of score tests for inference on variance components.

Whenever inference for variance components is required, the choice between one-sided and two-sided tests is crucial. This choice is usually driven by whether or not negative variance components are permitted. For two-sided tests, classical inferential procedures can be followed, based on likelihood ratios, score statistics, or Wald statistics. For one-sided tests, however, one-sided test statistics need to be developed, and their null distribution derived. While this has received considerable attention in the context of the likelihood ratio test, there appears to be much confusion about the related problem for the score test. The aim of this paper is to illustrate that classical (two-sided) score test statistics, frequently advocated in practice, cannot be used in this context, but that well-chosen one-sided counterparts could be used instead. The relation with likelihood ratio tests will be established, and all results are illustrated in an analysis of continuous longitudinal data using linear mixed models.

Analysis of Variance↗

PubMed2002

Impaired tolerance for glucose in women with recurrent vaginal candidiasis.

OBJECTIVE: The purpose of this study was to determine whether nondiabetic women with recurrent vaginal candidiasis have an impaired glucose metabolism. STUDY DESIGN: A case-control study of 62 otherwise healthy women who were attending a vaginitis clinic > or =3 times a year for symptoms of Candida vaginitis, positive microscopy, and at least one positive Candida culture and of 32 Candida-negative control subjects, all of whom were undergoing standardized oral glucose tolerance testing. RESULTS: Women with recurrent bacterial vaginal infections did not differ from women without infections, so both groups comprised the control group. Women with recurrent vaginal candidiasis had a greater mean body mass index than the control subjects (23.5 vs 21.4, P =.001). They had no more incidences of overt or preclinical diabetes mellitus than the control subjects (6/62 vs 0/32 incidents), but a greater proportion of them had at least one glucose concentration above the 95th percentile (36% vs 12%, P =.016). Glucose concentrations were higher in recurrent vaginal candidiasis cases than in control subjects at 0 (89 vs 85 mg/dL,P =.049), 30 (139 vs 126 mg/dL, P =.05), and 60 minutes (123 vs 102 mg/dL, P =.009) after the intake of 75 g of glucose. Fasting concentration of glycosylated hemoglobin was 25% higher in women with recurrent vaginal candidiasis (5 vs 4 g/dL, P =.0006), even after correction for body mass index. Compared with control subjects, ingestion of 75 g of glucose led to a 15% increase of serum glucose levels in women with recurrent vaginal candidiasis (P =.01). As expected, most of these differences were largely mediated by an increased body mass index. CONCLUSION: The tolerance to glucose in nondiabetic women with recurrent vaginal candidiasis is discretely impaired. Glucose tolerance testing is of value in women with recurrent vaginal candidiasis, but the interpretation of the obtained values should not be limited to the diagnosis of preclinical diabetes mellitus.

Adult↗

PubMed2002

Strategies to fit pattern-mixture models.

Whereas most models for incomplete longitudinal data are formulated within the selection model framework, pattern-mixture models have gained considerable interest in recent years (Little, 1993, 1994). In this paper, we outline several strategies to fit pattern-mixture models, including the so-called identifying restrictions strategy. Multiple imputation is used to apply this strategy to realistic settings, such as quality-of-life data from a longitudinal study on metastatic breast cancer patients.

Journal Article↗

PubMed2002

Group sequential methods for an ordinal logistic random-effects model under misspecification.

Interim analyses in clinical trials are planned for ethical as well as economic reasons. General results have been published in the literature that allow the use of standard group sequential methodology if one uses an efficient test statistic, e.g., when Wald-type statistics are used in random-effects models for ordinal longitudinal data. These models often assume that the random effects are normally distributed. However, this is not always the case. We will show that, when the random-effects distribution is misspecified in ordinal regression models, the joint distribution of the test statistics over the different interim analyses is still a multivariate normal distribution, but a sandwich-type correction to the covariance matrix is needed in order to obtain the correct covariance matrix. The independent increment structure is also investigated. A bias in estimation will occur due to the misspecification. However, we will also show that the treatment effect estimate will be unbiased under the null hypothesis, thus maintaining the type I error. Extensive simulations based on a toenail dermatophyte onychomycosis trial are used to illustrate our results.

Biometry↗