Linear mixed-effects models for central statistical monitoring of multicenter clinical trials.

Multicenter studies are widely used to meet accrual targets in clinical trials. Clinical data monitoring is required to ensure the quality and validity of the data gathered across centers. One approach to this end is central statistical monitoring, which aims at detecting atypical patterns in the data by means of statistical methods. In this context, we consider the simple case of a continuous variable, and we propose a detection procedure based on a linear mixed-effects model to detect location differences between each center and all other centers. We describe the performance of the procedure as a function of contamination rate and signal-to-noise ratio. We investigate the effect of center size and variance structure and illustrate the use of the procedure using data from two multicenter clinical trials.

Outcome of per protocol best-evidence based routine breast cancer care in a large regional hospital in Belgium: the importance of a prospective database in quality assurance.

AIM: Criteria for future accreditation of breast cancer centres in Belgium will be mainly based on the case load per surgeon or per centre. We would like to argue that the prospective collection of relevant data and the analysis of treatment related outcome derived from these data is feasible and should be the ultimate criterion for quality assessment and thus for accreditation since outcome is a more direct measurement of quality. METHODS: Data were prospectively collected on 715 invasive non metastatic breast cancers between 2002 and 2007 treated according to standard, best-evidence protocols in the setting of a large district hospital. Univariate and multivariate survival analysis were performed and compared to national and international databases. RESULTS: 5 year disease-free survival (DFS) and overall survival (OS) in our series were respectively 77 and 84%. In the multivariate analysis of DFS, only her-2-neu status (her-2-neu positivity being associated with a poor prognosis) and age (older age being a worse prognostic factor) were statistically significant prognostic factors. For OS, her-2-neu, age, and positive nodes were statistically significant prognostic factors. The outcome is comparable to other data sets. CONCLUSION: Centres dedicated to the care of women with breast cancer have the moral duty to produce outcome based results of their treatment. This report shows that such a collection of data is feasible and can be imposed as a prerequisite for accreditation. We also argue that outcome based data of treatment are a more solid base for quality assurance than case load.

Statistical Power Calculations for Clustered Continuous Data.

To calculate the sample size for a research study it is important to take into account several aspects of the study design. In particular, one needs to take into account the hypotheses being tested, the study design, the sampling design, and the method to be used for the analysis. In this paper we propose a simple method to calculate sample size for clustered continuous data under various scenarios of study design.

An investigation on performance of Significance Analysis of Microarray (SAM) for the comparisons of several treatments with one control in the presence of small-variance genes.

One of multiple testing problems in drug finding experiments is the comparison of several treatments with one control. In this paper we discuss a particular situation of such an experiment, i.e., a microarray setting, where the many-to-one comparisons need to be addressed for thousands of genes simultaneously. For a gene-specific analysis, Dunnett's single step procedure is considered within gene tests, while the FDR controlling procedures such as Significance Analysis of Microarrays (SAM) and Benjamini and Hochberg (BH) False Discovery Rate (FDR) adjustment are applied to control the error rate across genes. The method is applied to a microarray experiment with four treatment groups (three microarrays in each group) and 16,998 genes. Simulation studies are conducted to investigate the performance of the SAM method and the BH-FDR procedure with regard to controlling the FDR, and to investigate the effect of small-variance genes on the FDR in the SAM procedure.

Assessment of the usefulness of the fenestration method in cases of disc extrusion in the cervical and thoraco-lumbar spine in chondrodystrophic dogs.

The aim of this paper was to assess the effectiveness of treatment of Type I (according to Hansen) intervertebral disc extrusion using the fenestration method. The evaluated clinical material included 37 chondrodystrophic dogs in which multiple fenestration (from 2 to 6 intervertebral spaces) was performed. Dachshunds comprised 86.5% of patients. All dogs were administered corticosteroids during the operation. A mass similar to a slightly jelly-like yogurt, cottage cheese or plaster-like consistency was extracted. In one case a post-surgical pyogenic infection of the wound occured. In two cases, a deterioration in the neurological state followed the fenestration procedure of the cervical spine. In one case of thoraco-lumbar spine fenestration, a deterioration of clinical state was found. Recovery was observed after disc extrusion from the cervical spine in 8 out of 11 dogs, from the thoraco-lumbar spine in 12 out of 14 dogs with the 2nd grade clinical signs and in 11 out of 12 dogs with the 3rd grade clinical sings. In 6 dogs, recurrence of first or second grade clinical signs occurred, but only half of them had to undergo treatment. The probability of dog recovery did not depend on the degree of symptom intensity in a statistically significant way (p=0.11) or on the duration of the disease before the surgical treatment (p=0.87).

The assessment of the usefulness of hemilaminectomy without fenestration in the treatment of thoracolumbar disc disease in chondrodystrophic dogs.

The aim of the publication is to assess the effectiveness of the treatment of type I disc extrusion according to Hansen by using the method of hemilaminectomy without fenestration. The evaluated clinical material comprised 36 chondrodystrophic dogs in which 38 surgical procedures were performed. All dogs were administered corticosteroid during the operation. Dachshunds made up to 79% of patients. In 34 cases grainy mass was extracted from the vertebral canal. In two cases it was plaster-like mass and in other two--granulo-chondral. Most often (12 cases) the prolapse of the nucleus pulposus was observed between the first and second lumbar vertebra. There was one case of the postoperative pyogenic infection of the wound. Recovery was observed in all dogs with 3rd grade clinical signs, 17 out of 18 dogs with 4th grade and 8 out of 13 dogs with 5th grade of dysfunction severity. The recovery time of the dog depended in the statistically significant way on the severity of symptoms and the grade of the present disease (p<0.00005). There did not appear to be any statistically significat relationship between the recovery time and the duration of symptoms pre operatively (p=0.42), the time of non-ambulatory status (p=0.27) and the breed of the dog (p=0.81).

Risk factors for bovine herpesvirus-1 seropositivity.

This paper reports the investigation of risk factors for bovine herpesvirus-1-seropositivity, based on a cluster-sample survey of the Belgian cattle population. This serosurvey was carried out in 1998 in 309 randomly selected unvaccinated herds of all types (dairy, mixed and beef) were all bovids (N = 11,284) were sampled. Older and male cattle had higher seroprevalence. Origin (homebred or purchased) and herd size interacted; for smaller herds (< or = 50 cattle on the premises), purchase status and larger herd size were risk factors, whereas these effects were not observed for larger herds.

The effects of foot disease on quality of life: results of the Achilles Project.

BACKGROUND: Foot diseases have a high prevalence in the general population, but their impact on quality of life has not been assessed in large-scale studies. The Achilles Project surveyed foot disease in patients visiting their primary care physician or dermatologist. METHODS: A total of 43 593 patients were asked about the impact of their condition on their quality of life: pain, discomfort in walking, limitations in daily activities, and embarrassment. RESULTS: Overall, 52.5% of patients had some aspect of their quality of life affected by their foot disease. More specifically, 30.7% of patients experienced pain, 40.3% had discomfort in walking, 19.6% had their daily activities limited, and 27.3% were embarrassed. The survey indicated a larger impact of foot disease on the quality of life of women vs. men in all categories, except for daily activities. Similarly, the elderly (> or = 65 years) were more affected by their foot disease, although they suffered no more embarrassment than other age groups. Participation in sports seemed to lower the proportion of patients who had their quality of life adversely affected. Non-fungal foot diseases, particularly ulcer and gangrene, are more likely to cause pain, discomfort in walking and limit daily activities, than fungal diseases. CONCLUSIONS: In general, non-fungal foot diseases caused pain, discomfort in walking and limitations in daily activities in more patients than fungal foot diseases, but a higher proportion of patients with fungal foot diseases were embarrassed by their condition than patients with non-fungal foot diseases. The study found that the impact of foot disease on quality of life may be greater than previously suggested. Given that effective treatments are available, routine examination of patients' feet by dermatologists and primary care physicians may help to reduce the burden of these foot conditions.

High prevalence of foot diseases in Europe: results of the Achilles Project.

OBJECTIVE: To provide an insight into the prevalence of foot disease in Europe, and to include an assessment of the prevalence of predisposing factors and their correlation with foot disease. DESIGN: Large population-based survey conducted in 16 European countries. SETTING: The project consisted of two parts (study I and study II), in which all patients presenting to general practitioners and dermatologists over a defined time period were invited to participate. Patients. In study I, 70,497 patients presenting to dermatologists or general practitioners were recruited, and in study II 19,588 patients presenting to dermatologists were recruited. MAIN OUTCOME MEASURE: The feet of all participants were examined for signs of foot disease. The assessors also recorded relevant details such as the age and sex of patients, and the presence of predisposing factors for foot disease. In addition, patients in study II were offered a free mycological examination of the toenails and skin on the feet. RESULTS: In study I, 57.0% of patients had at least one foot disease. In study II, 61.3% had at least one foot disease. The proportions of patients with fungal foot disease and non-fungal foot disease in study I were 34.9% and 38.4%, respectively, and in study II were 40.6% and 41.7%, respectively. Orthopedic conditions and metatarsal corns were the most frequently reported non-fungal foot diseases, and onychomycosis and tinea pedis were the most frequently observed fungal infections. CONCLUSIONS: This large-scale survey suggests that the prevalence of fungal and non-fungal foot disease is higher than previously estimated.

Design-based analysis of surveys: a bovine herpesvirus 1 case study.

This paper critically assesses the design implications for the analysis of surveys of infections. It indicates the danger of not accounting for the study design in the statistical investigation of risk factors. A stratified design often implies an increased precision while clustering of infection results in a decreased precision. Through pseudo-likelihood estimation and linearisation of the variance estimator, the design effects can be taken into account in the analysis. The intra-cluster-correlation can be investigated through a logistic random effect model and a generalised estimating equation (GEE), allowing the investigation of the extent of spread of infections in a herd (cluster). The advantage of using adaptive Gaussian quadrature in a logistic random effect model is discussed. Applicable software is briefly reviewed. The methods are illustrated with data from a bovine herpesvirus 1 (BHV-1) serosurvey of Belgian cattle.

Expression of vascular endothelial growth factor (VEGF) in non-small cell lung cancer (NSCLC): association with p53 gene mutation and prognosis.

Vascular endothelial growth factor (VEGF) is a multifunctional cytokine that increases microvascular permeability and directly stimulates endothelial cell growth and angiogenesis. Recent evidence suggests that the genetic regulation of angiogenesis is also of crucial importance and that oncogenes and tumor suppressor genes can regulate it. The aim of this study was to determine the prognostic value of VEGF and its possible association with p53-gene mutation in 89 stage I-IIIa surgically treated NSCLC patients. DNA sequencing of the p53 gene (exons 5-8) showed 40 mutations (45%). Among the 89 NSCLC patients, immunoreactivity for VEGF was weakly, moderately and strongly positive in 35 (39%),36 (40%) and 18 (20%) cases, respectively. A strong, statistically significant association was found between the presence of a p53 gene mutation and expression of VEGF (P<0.001). The positive result of the p53 mutation increased the odds of observing a higher level of VEGF expression approximately 9.5 times (95% confidence interval: [3.44,25.89]). In the univariate analysis of survival, increasing levels of VEGF expression were associated with poor prognosis (P<0.001 for trend). In the multivariate analysis, after adjusting for the presence of a p53-gene mutation, gender, TNM stage and histological type, the prognostic effect of VEGF expression level was marginally non-significant (P=0.077). When the two-category quantification of the VEGF level was considered (low vs. intermediate/high), a marginally significant (P=0.024), unfavorable effect of intermediate/high levels of VEGF expression, independent of the effect of the presence of a p53-gene mutation, was found. In conclusion, we found that the p53 mutation was closely related to VEGF expression. Additionally, we observed that an intermediate/high expression of VEGF might be a useful indicator of prognosis in NSCLC. This latter conjecture, suggested by an analysis of the data, ought however, to be independently verified in further studies.

Prognostic significance of p53 and bcl-2 abnormalities in operable nonsmall cell lung cancer.

The association of p53 abnormalities and bcl-2 protein expression with clinical data and prognosis in 102 patients with resected nonsmall cell lung cancer (NSCLC) was investigated. Deoxyribonucleic acid analysis of exons 5-8 of the p53 gene showed mutations (p53-M) in 47% of resected NSCLC, serum p53 antibodies (p53-Abs) were detected in 25%, p53 protein overexpression (p53-PE) in 54%, and bcl-2 protein overexpression (bcl-2-PE) in 48%. A statistically significant association was found between p53-PE, serum p53-Abs and the presence of a p53 gene alteration. No significant associations were found between results of the p53-M, p53-Abs, bcl-2-PE tests and clinicopathological parameters. In the case of the p53-PE test there were significantly fewer positive results for adenocarcinoma than for squamous cell carcinoma and large cell carcinoma. Survival analysis showed that both p53 abnormalities and negative staining for bcl-2, when analysed separately, were associated with poor overall survival. In a multivariate analysis, only the positive result of the p53-M test remained an independent, statistically significant, unfavourable prognostic factor for survival. When the p53 mutation test was removed from the model, positive results of the p53-PE test and the p53-Abs test became statistically significant, unfavourable prognostic factors. To conclude, among p53 and bcl-2 abnormalities, only p53 gene mutations seem to have a strong and independent effect on prognosis. When deoxyribonucleic acid sequence information is not available, p53 protein expression and the presence of p53 antibodies in serum may be used to obtain important prognostic information.

Strong association between P53 protein accumulation, serum antibodies and gene mutation in non-small cell lung cancer.

DNA sequencing is the gold-standard method, but it is not feasible on a routine clinical basis. Immunohistochemistry is the most widely used method for assessing P53 alterations in human cancers. It can be performed during routine analysis, but some mutations may not lead to P53 protein accumulation, or P53 overexpression may be detected in the absence of mutations of the P53 gene. Recent studies have demonstrated the appearance of P53 antibodies in the serum of patients with lung cancer, probably due to the accumulation of mutant P53 protein in tumor cells. Using a logistic regression model applied to data for 102 non-small cell lung cancer (NSCLC) patients, we show a strong association between results of P53 mutation (P53-M) test, TNM stage and results of anti-P53 antibody in serum (P53-Abs) and P53 protein expression (P53-PE) tests. According to that model, we propose a procedure allowing for prediction of result of P53-M test. The percentage of correct predictions for independent data of 15 NSCLC patients was 80%. We conclude that in the absence of P53-M test result, a reasonably precise prediction of the test can be obtained using TNM stage and results of P53-Abs and P53-PE tests. The prediction can in turn be used to evaluate prognosis of NSCLC patients.

Relation between tumour response to first-line chemotherapy and survival in advanced colorectal cancer: a meta-analysis. Meta-Analysis Group in Cancer.

BACKGROUND: Treatment of advanced colorectal cancer has progressed substantially. However, improvements in response rates have not always translated into significant survival benefits. Doubts have therefore been raised about the usefulness of tumour response as a clinical endpoint. METHODS: This meta-analysis was done on individual data from 3791 patients enrolled in 25 randomised trials of first-line treatment with standard bolus intravenous fluoropyrimidines versus experimental treatments (fluorouracil plus leucovorin, fluorouracil plus methotrexate, fluorouracil continuous infusion, or hepatic-arterial infusion of floxuridine). Analyses were by intention to treat. FINDINGS: Compared with bolus fluoropyrimidines, experimental fluoropyrimidines led to significantly higher tumour response rates (454 responses among 2031 patients vs 209 among 1760; odds ratio 0.48 [95% CI 0.40-0.57], p<0.0001) and better survival (1808 deaths among 2031 vs 1580 among 1760; hazard ratio 0.90 [0.84-0.97], p=0.003). The survival benefits could be explained by the higher tumour response rates. However, a treatment that lowered the odds of failure to respond by 50% would be expected to decrease the odds of death by only 6%. In addition, less than half of the variability of the survival benefits in the 25 trials could be explained by the variability of the response benefits in these trials. INTERPRETATION: These analyses confirm that an increase in tumour response rate translates into an increase in overall survival for patients with advanced colorectal cancer. However, in the context of individual trials, knowledge that a treatment has benefits on tumour response does not allow accurate prediction of the ultimate benefit on survival.

p53 gene mutation and protein expression in operable non-small cell lung cancer in Poland.

We investigated the association of p53 abnormalities (gene mutations by DNA sequencing and protein over-expression by immunostaining) with clinical data and prognosis in 74 patients with resected non-small cell lung cancer (NSCLC). DNA analysis of exons 5-8 of the p53 gene showed 34 mutations in 74 resected primary NSCLC (45.9%). Immunohistochemical study of the p53 protein revealed that 41 of 74 (55.4%) samples had positive staining. We found strong agreement between the results of the p53 protein expression test (p53-PE) and the p53 gene mutation test (p53-M) (Cohen's kappa = 0.65, 95% CI 0.48-0.82). Joint distribution of the results (analysed using the bivariate Dale model) was mainly influenced by, histological type of tumour. A positive result for the p53-PE test significantly increased (estimated odds ratio 84.5; 95% CI 8.89-803.03) the odds of observing a positive result in the p53-M test. In the univariate analysis (log rank test), positive results in the p53-M test and the p53-PE test were significantly associated with overall survival (P < 0.001 and P = 0.005, respectively). In the multivariate analysis (Cox's proportional hazard model), a positive result for the p53-M test significantly increased relative risk for overall survival (RR 9.56; 95% CI 2.62-34.87; P < 0.001). When the result of the p53-M test was accounted for, a positive result for the p53-PE test did not offer any additional prognostic information due to the strong dependence of results of the tests. However, when the result of the p53-M test was removed from the model, a positive result for the p53-PE test became a significant unfavourable prognostic factor (P = 0.009). We conclude that p53 gene mutation and protein expression analyses are in a strong agreement. Joint distribution of the results depends mainly on histological type of tumour. When considered separately, both tests are unfavourable prognostic factors in NSCLC. When the result of the p53-M test is taken into account, the p53-PE test does not offer any additional prognostic information.

The validation of surrogate endpoints in meta-analyses of randomized experiments.

The validation of surrogate endpoints has been studied by Prentice (1989). He presented a definition as well as a set of criteria, which are equivalent only if the surrogate and true endpoints are binary. Freedman et al. (1992) supplemented these criteria with the so-called 'proportion explained'. Buyse and Molenberghs (1998) proposed replacing the proportion explained by two quantities: (1) the relative effect linking the effect of treatment on both endpoints and (2) an individual-level measure of agreement between both endpoints. The latter quantity carries over when data are available on several randomized trials, while the former can be extended to be a trial-level measure of agreement between the effects of treatment of both endpoints. This approach suggests a new method for the validation of surrogate endpoints, and naturally leads to the prediction of the effect of treatment upon the true endpoint, given its observed effect upon the surrogate endpoint. These ideas are illustrated using data from two sets of multicenter trials: one comparing chemotherapy regimens for patients with advanced ovarian cancer, the other comparing interferon-alpha with placebo for patients with age-related macular degeneration.

The effect of long-term water storage on the tensile strength of orthodontic brackets bonded with resin-reinforced glass-ionomer cements.

The purpose of this study was to evaluate in vitro the effect of water storage on the tensile bond strength of orthodontic brackets bonded with Vitremer and Fuji II LC resin-reinforced glass-ionomer cements. Seventy-two extracted human premolars were randomly divided into 6 groups and the bonding strengths of the resin-reinforced glass-ionomer cements were compared to control groups bonded with Concise composite resin at 24 hours and 9 months. The brackets were bonded on prepared teeth and a tensile load was applied to dislodge the brackets held in a special device from Lloyd 1000R testing machine. The effects of duration of water exposure, type of bonding material and interaction between long-term water exposure and type of bonding material on the bonding strength were described using the Weibull regression model. The mean tensile bond strength of resin-reinforced glass-ionomers after 24 hours water storage was significantly lower than the mean of the control samples. The results of this study showed no effect of water exposure on the tensile bond strength of brackets bonded with Vitremer glass-ionomer cement. The mean bonding strength of brackets bonded with Fuji II LC without enamel pre-conditioning significantly increased after 9-months water storage as compared to Fuji samples exposed to water for 24 hours. The results of this investigation allow to conclude: 1. Long-term water storage had not decreased the bond strength of resin-reinforced glass-ionomers applied as orthodontic adhesives, 2. Weibull regression model is an adequate and flexible tool to evaluate the bonding properties of dental materials.

Low-dose glucose-insulin-potassium is ineffective in acute myocardial infarction: results of a randomized multicenter Pol-GIK trial.

We aimed to assess the clinical efficacy of glucose-insulin-potassium (GIK) in acute myocardial infarction. Experimental data provided evidence of the beneficial effects of GIK on ischemic myocardium. The clinical trials, mostly uncontrolled and conducted mainly before the thrombolytic era, were inconclusive due to the small number of patients and discrepancies in protocols. In order to evaluate the efficacy of this intervention, we have performed a prospective multicenter randomized study. The study consisted of 954 patients with acute myocardial infarction (MI) randomized within 24 hours from the onset of symptoms to low-dose GIK (n = 494), which consisted of 1000 mL 10% dextrose, 32-20 U insulin, and 80 mEq K-, or to the control group (n = 460), which was given 1000 mL 0.89% sodium chloride, by intravenous 24-hour infusion at a rate of 42 mL/h. Cardiac mortality and the occurrence of cardiac events at 35 days did not differ between GIK and control-allocated patients (32 (6.5%) vs. 21 (4.6%), respectively; OR 1.45, 95% CI 0.79-2.68, P = 0.20; and 214 (43.3%) vs. 192 (41.7%), OR 1.07, 95% CI 0.82-1.38, P = 0.62). Total mortality at 35 days was significantly higher in the GIK than in the control group (44 (8.9%) vs. 22 (4.8%), respectively, OR 1.95, 95% CI 1.12-3.47, P = 0.01). The excess of non-cardiac deaths in the GIK group may have occurred by chance. Low-dose GIK treatment does not improve the survival and clinical course in acute MI.

Analysis of incomplete public health data.

The problem of dealing with missing values is common throughout statistics and is very prominent with epidemiologic data in the broad sense. Not only do data collection procedures break down, but subjects may be lost to follow up, or simply withdraw their consent without further providing a reason for doing so. In this paper, we review a framework for handling incomplete studies, and then concentrate on a specific case. It comes from a complex health interview survey, conducted in Belgium in 1997, where different types of missingness arise at various levels of the hierarchical sampling procedure.