To condition or not condition? Analysing 'change' in longitudinal randomised controlled trials.

OBJECTIVE: The statistical analysis for a 2-arm randomised controlled trial (RCT) with a baseline outcome followed by a few assessments at fixed follow-up times typically invokes traditional analytic methods (eg, analysis of covariance (ANCOVA), longitudinal data analysis (LDA)). 'Constrained' longitudinal data analysis (cLDA) is a well-established unconditional technique that constrains means of baseline to be equal between arms. We use an analysis of fasting lipid profiles from the Group Medical Clinics (GMC) longitudinal RCT on patients with diabetes to illustrate applications of ANCOVA, LDA and cLDA to demonstrate theoretical concepts of these methods including the impact of missing data. METHODS: For the analysis of the illustrated example, all models were fit using linear mixed models to participants with only complete data and to participants using all available data. RESULTS: With complete data (n=195), 95% CI coverage are equivalent for ANCOVA and cLDA with an estimated 11.2 mg/dL (95% CI -19.2 to -3.3; p=0.006) lower mean low-density lipoprotein (LDL) cholesterol in GMC compared with usual care. With all available data (n=233), applying the cLDA model yielded an LDL improvement of 8.9 mg/dL (95% CI -16.7 to -1.0; p=0.03) for GMC compared with usual care. The less efficient, LDA analysis yielded an LDL improvement of 7.2 mg/dL (95% CI -17.2 to 2.8; p=0.15) for GMC compared with usual care. CONCLUSIONS: Under reasonable missing data assumptions, cLDA will yield efficient treatment effect estimates and robust inferential statistics. It may be regarded as the method of choice over ANCOVA and LDA.

Slope estimation for informatively right censored longitudinal data modelling the number of observations using geometric and Poisson distributions: application to renal transplant cohort.

Analysis of longitudinal data is often complicated by the presence of informative right censoring. This type of censoring should be accounted for in the analysis so that valid slope estimates are attained. In this study, we developed a new likelihood-based approach wherein the likelihood function is integrated over random effects to obtain a marginal likelihood function. Maximum likelihood estimates for the population slope were acquired by direct maximisation of the marginal likelihood function and empirical Bayes estimates for the individual slopes were generated using Gaussian quadrature. The performance of the model was assessed using the geometric and Poisson distributions to model the number of observations for every individual subject. Our model generated valid estimates for the slopes under both distributions with minimal bias and mean squared errors. Our sensitivity analysis confirmed the robustness of the model to assumptions pertaining to the underlying distribution and demonstrated its insensitivity to normality assumptions. Moreover, superiority of the model in terms of accuracy of slope estimates was consistently shown across the different levels of censoring in comparison to the naïve and bootstrap approaches. This model was illustrated using the cohort of renal transplant patients and estimates of the slopes that are adjusted for informative right censoring were acquired.

Challenges of decision making regarding futility in a randomized trial: the Interventional Management of Stroke III experience.

BACKGROUND AND PURPOSE: Interventional Management of Stroke (IMS) III is a randomized, parallel arm trial comparing the approach of intravenous tissue-type plasminogen activator followed by endovascular treatment with intravenous tissue-type plasminogen activator alone in patients with acute ischemic stroke presenting <3 hours of symptom onset. The trial intended to enroll 900 subjects to ensure adequate statistical power to detect an absolute 10% difference in the percentage of subjects with good outcome, defined as modified Rankin Scale score of 0 to 2 at 3 months. In April 2012, after 656 subjects were randomized, further enrollment was terminated by the National Institute of Neurological Disorders and Stroke based on the prespecified criterion for futility using conditional power<20%. METHODS: Conditional power was defined as the likelihood of finding statistical significance at the end of the study, given the accumulated data to date and with the assumption that a minimum hypothesized difference of 10% truly exists between the 2 groups. The evolution of study data leading to futility determination is described, including the interaction between the unblinded study statisticians and the Data and Safety Monitoring Board in the complex deliberation of analysis results. RESULTS: The futility boundary was crossed at the trial's fourth interim analysis. At this point, based on the conditional power criteria, the Data and Safety Monitoring Board recommended termination of the trial. CONCLUSIONS: Even in spite of prespecified interim analysis boundaries, interim looks at data pose challenges in interpretation and decision making, underscoring the importance of objective stopping criteria. CLINICAL TRIAL REGISTRATION URL: http://www.clinicaltrials.gov. Unique identifier: NCT00359424.

Move over LOCF: principled methods for handling missing data in sleep disorder trials.

Missing data, e.g. patient attrition, are endemic in sleep disorder clinical trials. Common approaches for dealing with this situation include complete-case analysis (CCA) and last observation carried forward (LOCF). Although these methods are simple to implement, they are deeply flawed in that they may introduce bias and underestimate uncertainty, leading to erroneous conclusions. There are alternative principled approaches, however, that are available in statistical software namely mixed-effects models and multiple imputation. In this paper we introduce terminology used to describe different assumptions about missing data. We emphasize that understanding reasons for missingness is a critical step in the analysis process. We describe and implement both linear mixed-effects models and an inclusive multiple imputation strategy for handling missing data in a randomized trial examining sleep outcomes. These principled strategies are compared with "complete-case analysis" and LOCF. These analyses illustrate that methodologies for accommodating missing data can produce different results in both direction and strength of treatment effects. Our goal is for this paper to serve as a guide to sleep disorder clinical trial researchers on how to utilize principled methods for incomplete data in their trial analyses.

Application of beta regression to analyze ischemic stroke volume in NINDS rt-PA clinical trials.

BACKGROUND AND PURPOSE: Ischemic stroke lesion volumes have proven difficult to analyze due to the extremely skewed shape of their underlying distribution. We introduce an extension of generalized linear models, beta regression, as a possible method of modeling extremely skewed distributions as evidenced in ischemic stroke lesion volumes. METHODS: The NINDS rt-PA clinical trials measured ischemic stroke lesion volume as a secondary trial outcome. Three-month lesion volumes from these trials were analyzed using beta regression. A multi-variable regression model associating explanatory variables with ischemic stroke lesion volumes was constructed using accepted model building strategies and compared with the previously published volumetric analysis. RESULTS: Beta regression produced a similar model when compared to the previous analysis published by the study group. All previously identified variables of importance were detected in the model building process. The age by treatment interaction described in previous studies was also found in this analysis, confirming the strong effect age has on stroke outcomes. Further, a treatment effect was elicited in terms of odds ratios, yielding a previously unknown quantification of the effect of rt-PA on lesion volumes. CONCLUSIONS: Beta regression proved adept in modeling ischemic stroke lesions and offered the interpretation of covariates in terms of odds ratios. Beta regression is seen as a legitimate alternative to analyze ischemic stroke volumes.

Measuring blood pressure for decision making and quality reporting: where and how many measures?

BACKGROUND: The optimal setting and number of blood pressure (BP) measurements that should be used for clinical decision making and quality reporting are uncertain. OBJECTIVE: To compare strategies for home or clinic BP measurement and their effect on classifying patients as having BP that was in or out of control. DESIGN: Secondary analysis of a randomized, controlled trial of strategies to improve hypertension management. (ClinicalTrials.gov registration number: NCT00237692) SETTING: Primary care clinics affiliated with the Durham Veterans Affairs Medical Center. PATIENTS: 444 veterans with hypertension followed for 18 months. MEASUREMENTS: Blood pressure was measured repeatedly by using 3 methods: standardized research BP measurements at 6-month intervals; clinic BP measurements obtained during outpatient visits; and home BP measurements using a monitor that transmitted measurements electronically. RESULTS: Patients provided 111,181 systolic BP (SBP) measurements (3218 research, 7121 clinic, and 100,842 home measurements) over 18 months. Systolic BP control rates at baseline (mean SBP<140 mm Hg for clinic or research measurement; <135 mm Hg for home measurement) varied substantially, with 28% classified as in control by clinic measurement, 47% by home measurement, and 68% by research measurement. Short-term variability was large and similar across all 3 methods of measurement, with a mean within-patient coefficient of variation of 10% (range, 1% to 24%). Patients could not be classified as having BP that was in or out of control with 80% certainty on the basis of a single clinic SBP measurement from 120 mm Hg to 157 mm Hg. The effect of within-patient variability could be greatly reduced by averaging several measurements, with most benefit accrued at 5 to 6 measurements. LIMITATION: The sample was mostly men with a long-standing history of hypertension and was selected on the basis of previous poor BP control. CONCLUSION: Physicians who want to have 80% or more certainty that they are correctly classifying patients' BP control should use the average of several measurements. Hypertension quality metrics based on a single clinic measurement potentially misclassify a large proportion of patients. PRIMARY FUNDING SOURCE: U.S. Department of Veterans Affairs Health Services Research and Development Service.

Slope Estimation for Bivariate Longitudinal Outcomes Adjusting for Informative Right Censoring Using Discrete Survival Model: Application to the Renal Transplant Cohort.

Patients undergoing renal transplantation are prone to graft failure which causes lost of follow-up measures on their blood urea nitrogen and serum creatinine levels. These two outcomes are measured repeatedly over time to assess renal function following transplantation. Loss of follow-up on these bivariate measures results in informative right censoring, a common problem in longitudinal data that should be adjusted for so that valid estimates are obtained. In this study, we propose a bivariate model that jointly models these two longitudinal correlated outcomes and generates population and individual slopes adjusting for informative right censoring using a discrete survival approach. The proposed approach is applied to the clinical dataset of patients who had undergone renal transplantation. A simulation study validates the effectiveness of the approach.

Analyses of renal outcome following transplantation adjusting for informative right censoring and demographic factors: A longitudinal study.

Demographic factors such as race, vital status, gender, and age could affect the final renal outcome of patients who undergo renal transplantation. These demographic factors could be assessed at the recipient and donor levels. Repeated measures for blood urea nitrogen (BUN) are typically recorded for each patient following renal transplantation, as a biomarker to assess renal progress. However, once a patient develops renal failure due to graft rejection, no measurement of BUN can be registered and the patient goes back to dialysis. This causes loss of follow-up and incomplete data on BUN measurements, a problem referred to as informative right censoring. If this problem is ignored, inaccurate, and biased estimates will be generated. In this study, unbiased estimates for the rate of change of BUN levels over time adjusted for informative right censoring and demographic factors were acquired using a sophisticated model of analysis. Our results demonstrated that BUN levels for Caucasians were decreasing at a greater rate than African Americans (p < 0.0001). When donors are deceased, African American recipients showed an increase instead of a decrease in their BUN levels following transplantation. Moreover, African Americans showed a decrease in their BUN levels when the donors were African Americans compared with when donors were Caucasians (p = 0.03). Our results also showed that BUN levels were decreasing at a greater rate when donors and recipients were of different gender than when they were of the same gender (p = 0.009). These results suggest that the success of renal transplantation is impacted by the donor/recipient demographic factors.

Does dark chocolate have a role in the prevention and management of hypertension?: commentary on the evidence.

The notion that eating chocolate would prevent or treat hypertension is appealing to many who produce and enjoy chocolate. Several studies have documented beneficial effects of dark chocolate on insulin action and endothelial function. However, the published studies on chocolate and blood pressure include a relatively small number of subjects, and results are conflicting. In addition, because of secrecy surrounding the production of chocolate and the unique sociocultural context of this popular food, research on efficacy and effectiveness is complex. This commentary summarizes 13 peer-reviewed studies on dark chocolate and blood pressure and raises questions relevant to its future as an evidence-based lifestyle intervention.

Prevalence of hypertension by duration and age at exposure to the stroke belt.

Geographic variation in hypertension is hypothesized as contributing to the stroke belt, an area in the southeastern United States with high stroke mortality. No study has examined hypertension by lifetime exposure to the stroke belt. This association was studied in 19,385 participants in the REasons for Geographic And Racial Differences in Stroke (REGARDS) study, a national population-based cohort. Prevalent hypertension was defined as systolic blood pressure >/=140, diastolic blood pressure >/=90, or use of antihypertensive medications. Stroke belt exposure was assessed by residence at birth, currently, early childhood, adolescence, early adulthood, mid-adulthood, and recently. After adjustment for age, race, sex, physical activity level, body mass index, smoking, alcohol, education, and income, the prevalence of hypertension was significantly more strongly related (P < .0001) with lifetime exposure, adolescence, or early adulthood exposure than exposures at other times. Birthplace and current residence were independently associated with hypertension; however, lifetime, adolescence, or early adulthood exposures were more predictive than joint model with both birthplace and current residence. That adolescence and early adulthood periods are more predictive than residence in the stroke belt for most recent 20-year period suggests community and environmental strategies to prevent hypertension need to start earlier in life.

Using global statistical tests in long-term Parkinson's disease clinical trials.

Parkinson's disease (PD) impairments are multidimensional, making it difficult to choose a single primary outcome when evaluating treatments to stop or lessen the long-term decline in PD. We review commonly used multivariate statistical methods for assessing a treatment's global impact, and we highlight the novel Global Statistical Test (GST) methodology. We compare the GST to other multivariate approaches using data from two PD trials. In one trial where the treatment showed consistent improvement on all primary and secondary outcomes, the GST was more powerful than other methods in demonstrating significant improvement. In the trial where treatment induced both improvement and deterioration in key outcomes, the GST failed to demonstrate statistical evidence even though other techniques showed significant improvement. Based on the statistical properties of the GST and its relevance to overall treatment benefit, the GST appears particularly well suited for a disease like PD where disability and impairment reflect dysfunction of diverse brain systems and where both disease and treatment side effects impact quality of life. In future long term trials, use of GST for primary statistical analysis would allow the assessment of clinically relevant outcomes rather than the artificial selection of a single primary outcome.

The impact of loss to follow-up on hypothesis tests of the treatment effect for several statistical methods in substance abuse clinical trials.

"Loss to follow-up" can be substantial in substance abuse clinical trials. When extensive losses to follow-up occur, one must cautiously analyze and interpret the findings of a research study. Aims of this project were to introduce the types of missing data mechanisms and describe several methods for analyzing data with loss to follow-up. Furthermore, a simulation study compared Type I error and power of several methods when missing data amount and mechanism varies. Methods compared were the following: Last observation carried forward (LOCF), multiple imputation (MI), modified stratified summary statistics (SSS), and mixed effects models. Results demonstrated nominal Type I error for all methods; power was high for all methods except LOCF. Mixed effect model, modified SSS, and MI are generally recommended for use; however, many methods require that the data are missing at random or missing completely at random (i.e., "ignorable"). If the missing data are presumed to be nonignorable, a sensitivity analysis is recommended.

A comparison of missing data methods for hypothesis tests of the treatment effect in substance abuse clinical trials: a Monte-Carlo simulation study.

BACKGROUND: Missing data due to attrition are rampant in substance abuse clinical trials. However, missing data are often ignored in the presentation of substance abuse clinical trials. This paper demonstrates missing data methods which may be used for hypothesis testing. METHODS: Methods involving stratifying and weighting individuals based on missing data pattern are shown to produce tests that are robust to missing data mechanisms in terms of Type I error and power. In this article, we describe several methods of combining data that may be used for testing hypotheses of the treatment effect. Furthermore, illustrations of each test's Type I error and power under different missing data percentages and mechanisms are quantified using a Monte-Carlo simulation study. RESULTS: Type I error rates were similar for each method, while powers depended on missing data assumptions. Specifically, power was greatest for the weighted, compared to un-weighted methods, especially for greater missing data percentages. CONCLUSION: Results of this study as well as extant literature demonstrate the need for standards of design and analysis specific to substance abuse clinical trials. Given the known substantial attrition rates and concern for the missing data mechanism in substance abuse clinical trials, investigators need to incorporate missing data methods a priori. That is, missing data methods should be specified at the outset of the study and not after the data have been collected.

Effect of psyllium fiber supplementation on C-reactive protein: the trial to reduce inflammatory markers (TRIM).

PURPOSE: Recent evidence supports a significant association between the intake of dietary fiber and levels of inflammatory markers. The objective of this study was to determine whether daily fiber supplementation would reduce levels of inflammatory markers. METHODS: This study was a prospective randomized controlled trial at a single university medical center. Participants were overweight or obese adults with no history of heart disease. The intervention was psyllium supplementation at either 7 or 14 g/d for 3 months compared with no supplements in a control group. The main outcome measure was change in level of high-sensitivity C-reactive protein (hsCRP) concentration; secondary outcomes included changes in interleukin-6 (IL-6) levels, fibrinogen levels, and white blood cell (WBC) count. Protocol completers attended at least 2 visits and took more than 75% of the prescribed fiber dose. RESULTS: In this intent-to-treat analysis (n = 158), there were no significant differences between either of the 2 treatment groups and the control group in the amount of change in CRP, fibrinogen, or IL-6 levels or in WBC count (P>.05). In the analysis of protocol completers (n = 132), there also were no significant differences between the groups except for a small decrease in fibrinogen level in the high-fiber group (-6 mg/dL [-0.18 micromol/L] compared with 13 mg/dL [0.38 micromol/L] in the control group, P<.05). CONCLUSION: Psyllium fiber supplementation did not significantly reduce CRP levels in overweight or obese individuals in this trial, and changes in other markers were not consistent. Further research is needed to determine whether other fibers or nutrients can reduce inflammatory markers.

A rank-based sample size method for multiple outcomes in clinical trials.

O'Brien (Biometrics 1984; 40:1079-1087) introduced a rank-sum-type global statistical test to summarize treatment's effect on multiple outcomes and to determine whether a treatment is better than others. This paper presents a sample size computation method for clinical trial design with multiple primary outcomes, and O'Brien's test or its modified test (Biometrics 2005; 61:532-539) is used for the primary analysis. A new measure, the global treatment effect (GTE), is introduced to summarize treatment's efficacy from multiple primary outcomes. Computation of the GTE under various settings is provided. Sample size methods are presented based on prespecified GTE both when pilot data are available and when no pilot data are available. The optimal randomization ratio is given for both cases. We compare our sample size method with the Bonferroni adjustment for multiple tests. Since ranks are used in our derivation, sample size formulas derived here are invariant to any monotone transformation of the data and are robust to outliers and skewed distributions. When all outcomes are binary, we show how sample size is affected by the success probabilities of outcomes. Simulation shows that these sample size formulas provide good control of type I error and statistical power. An application to a Parkinson's disease clinical trial design is demonstrated. Splus codes to compute sample size and the test statistic are provided.

Differences in cardiovascular disease mortality associated with body mass between Black and White persons.

We analyzed cardiovascular disease mortality risks associated with obesity using participant-level meta-analysis of data from the Black Pooling Project for Black and White individuals. The adjusted relative risks (ARRs) were stronger among White participants than among Black participants for coronary heart disease AAR=1.21 (95% confidence interval [CI]=1.07, 1.36) versus 0.87 (95% CI=0.69, 1.09), respectively, and cardiovascular disease ARR=1.18 (95% CI=1.07, 1.29) versus 0.91 (95% CI=0.77, 1.05), respectively. The results suggest that obesity is an independent risk factor in White people, and additional study of body size and disease progression is necessary in the assessment of racial disparities.

Oral cancer prevention and early detection: using the PRECEDE-PROCEED framework to guide the training of health professional students.

BACKGROUND: Teaching cancer prevention and detection is important in health professional education. It is desirable to select a comprehensive framework for teaching oral cancer (OC) prevention and detection skills. METHODS: The PRECEDE-PROCEED model was used to design a randomized pretest and posttest study of the OC prevention and detection skills of dental students (n = 104). OC knowledge, opinions, and competencies were evaluated. RESULTS: Second year students in the intervention group were more competent than those in the control group. CONCLUSIONS: The novel use of PRECEDE-PROCEED sets a precedent for designing a standardized OC curriculum for a wide range of health professional disciplines.

Health services use among Gulf War veterans and Gulf War era nondeployed veterans: a large population-based survey.

We sought to analyze the self-reported hospitalization, emergency department visits, and outpatient visits of Persian Gulf War (deployed; n=1896) and Persian Gulf War-era (nondeployed; n=1799) military personnel 5 years postconflict to determine whether these groups had different rates of health care use. Compared with personnel who had not been deployed, personnel who had been deployed were more likely to have visited an emergency department (25% vs 21%; odds ratio [OR]=1.24; 95% confidence interval [CI] = 1.06, 1.51]). Among these groups, the National Guard and Reserve personnel were more likely to have been hospitalized than were the regular military personnel (OR= 1.65; 95% CI=1.21, 2.26).

Nephrogenic systemic fibrosis after exposure to gadolinium in patients with renal failure.

BACKGROUND: Nephrogenic systemic fibrosis is a debilitating disease occurring exclusively in patients with renal failure. The aetiology of nephrogenic systemic fibrosis is unclear, but recent reports suggest that exposure to gadolinium for enhancement of magnetic resonance imaging may play a role. In the present study, we assessed the association of exposure to gadolinium with the development of nephrogenic systemic fibrosis in patients with various stages of chronic kidney disease. METHODS: We analysed the exposure to gadolinium and development of nephrogenic systemic fibrosis in 849 patients on renal replacement therapy over 5 years. We also performed inquiry of development of the nephrogenic systemic fibrosis in 592 patients exposed to gadolinium and estimated to be in stages 3 and 4 of chronic kidney disease. RESULTS: In 849 patients undergoing chronic dialysis from 2001 through 2006 time period, four of the 261 who had received gadolinium (1.5%) and none of the 588 not exposed to gadolinium developed clinically apparent disease. The odds ratio for developing nephrogenic systemic fibrosis was 6.671 [95% confidence interval (CI) 1.537-53.97] in patients with a single gadolinium exposure compared to patients without gadolinium exposure. This ratio increased to 44.5 (95% CI 2.362-2913) in patients with multiple gadolinium exposures compared to patients not receiving gadolinium. None of the 592 patients estimated to be in stage 3 or 4 of chronic kidney disease developed nephrogenic systemic fibrosis after exposure to gadolinium. CONCLUSION: Gadolinium exposure is associated with nephrogenic systemic fibrosis in patients on chronic renal replacement therapy at a low rate. This association appears to increase with repeated exposure to gadolinium. Since nephrogenic systemic fibrosis may be clinically occult, its prevalence may be higher than reported. Despite this association, it is unclear if gadolinium is the sole or most important factor in the pathogenesis of the disease.

Use of the OSCE to evaluate brief communication skills training for dental students.

Although communications competency is recommended by the American Dental Education Association, only a few (n=5) dental schools report evaluating students' skills using a competency examination for communication. This study used an objective structured clinical examination (OSCE) to evaluate dental students' competency in interpersonal and tobacco cessation communication skills. All students were evaluated on their interpersonal communication skills at baseline and at six months post-OSCE by standardized patients and on their tobacco cessation communication skills by two independent raters. First- and second-year dental students (n=104) were randomized to a control or intervention group. One month after the baseline OSCE, students in the intervention group participated in a two-hour training session in which faculty members communicated with a standardized patient during a head and neck examination and counseled the patient about tobacco cessation. There were no statistically significant differences from baseline to post-test between the intervention and control group students as measured by the OSCE. However, among first-year students, both the intervention (n=23) and control (n=21) groups significantly increased in tobacco cessation communication scores. Second-year students in both intervention (n=24) and control (n=28) groups declined in interpersonal communication skills from baseline to post-test. Overall, this one-shot intervention was not successful, and results suggest that a comprehensive communication skills training course may be more beneficial than a single, brief training session for improving dental students' communication skills.