Discussion of "Optimal test procedures for multiple hypotheses controlling the familywise expected loss" by Willi Maurer, Frank Bretz, and Xiaolei Xun.

We provide commentary on the paper by Willi Maurer, Frank Bretz, and Xiaolei Xun entitled, "Optimal test procedures for multiple hypotheses controlling for the familywise expected loss." The authors provide an excellent discussion of the multiplicity problem in clinical trials and propose a novel approach based on a decision-theoretic framework that incorporates loss functions that can vary across multiple hypotheses in a family. We provide some considerations for the practical use of the authors' proposed methods as well as some alternative methods that may also be of interest in this setting.

Biomarker Qualification: Toward a Multiple Stakeholder Framework for Biomarker Development, Regulatory Acceptance, and Utilization.

The discovery, development, and use of biomarkers for a variety of drug development purposes are areas of tremendous interest and need. Biomarkers can become accepted for use through submission of biomarker data during the drug approval process. Another emerging pathway for acceptance of biomarkers is via the biomarker qualification program developed by the Center for Drug Evaluation and Research (CDER, US Food and Drug Administration). Evidentiary standards are needed to develop and evaluate various types of biomarkers for their intended use and multiple stakeholders, including academia, industry, government, and consortia must work together to help develop this evidence. The article describes various types of biomarkers that can be useful in drug development and evidentiary considerations that are important for qualification. A path forward for coordinating efforts to identify and explore needed biomarkers is proposed for consideration.

The role of nonrandomized trials in the evaluation of oncology drugs.

Although randomized trials provide the most reliable evidence of a drug's safety and efficacy, there are situations where randomized trials are not possible or ethical. In this article we discuss when and how single-arm trials can be used to support full approval of oncology drugs. These include situations in which an unprecedented effect on tumor response is observed in a setting of high unmet medical need, clinical trial patients have been well characterized, enabling a target population to be clearly defined, experience exists in a sufficient number of patients to allow adequate assessment of the risk:benefit relationship, and a proper historical context can be provided for analysis. We also discuss how response rates might be considered predictive of long-term outcomes or clinically meaningful in and of themselves in certain contexts.

Double-masked, placebo-controlled safety and efficacy trial of diquafosol tetrasodium (INS365) ophthalmic solution for the treatment of dry eye.

PURPOSE: To investigate the safety and efficacy of diquafosol tetrasodium, a P2Y2 receptor agonist that stimulates fluid and mucin secretion on the ocular surface, as a novel topical treatment of dry eye disease. METHODS: Subjects with dry eye (n=527) were evaluated in a randomized, double-masked, parallel-group trial comparing 24 weeks of treatment with 2 concentrations of diquafosol (1% and 2%) versus placebo instilled 4 times daily. Corneal staining, conjunctival staining, Schirmer tests, and subjective symptoms of dry eye were evaluated. Use of artificial tears was permitted as necessary. RESULTS: Subjects treated with 2% diquafosol had significantly lower corneal staining scores compared with placebo at the 6-week, primary efficacy time point (P<0.001), and superiority continued throughout the 24-week study. Reductions in corneal staining were observed as early as after 2 weeks of treatment, were maintained throughout the 24-week study, and were observed to worsen slightly (toward baseline) when diquafosol treatment was discontinued (week 25). Results for conjunctival staining were consistent with those observed for corneal staining. Schirmer scores at week 6 were significantly higher with diquafosol treatment than with placebo (P

Applying sample survey methods to clinical trials data.

This paper outlines the utility of statistical methods for sample surveys in analysing clinical trials data. Sample survey statisticians face a variety of complex data analysis issues deriving from the use of multi-stage probability sampling from finite populations. One such issue is that of clustering of observations at the various stages of sampling. Survey data analysis approaches developed to accommodate clustering in the sample design have more general application to clinical studies in which repeated measures structures are encountered. Situations where these methods are of interest include multi-visit studies where responses are observed at two or more time points for each patient, multi-period cross-over studies, and epidemiological studies for repeated occurrences of adverse events or illnesses. We describe statistical procedures for fitting multiple regression models to sample survey data that are more effective for repeated measures studies with complicated data structures than the more traditional approaches of multivariate repeated measures analysis. In this setting, one can specify a primary sampling unit within which repeated measures have intraclass correlation. This intraclass correlation is taken into account by sample survey regression methods through robust estimates of the standard errors of the regression coefficients. Regression estimates are obtained from model fitting estimation equations which ignore the correlation structure of the data (that is, computing procedures which assume that all observational units are independent or are from simple random samples). The analytic approach is straightforward to apply with logistic models for dichotomous data, proportional odds models for ordinal data, and linear models for continuously scaled data, and results are interpretable in terms of population average parameters. Through the features summarized here, the sample survey regression methods have many similarities to the broader family of methods based on generalized estimating equations (GEE). Sample survey methods for the analysis of time-to-event data have more recently been developed and implemented in the context of finite probability sampling. Given the importance of survival endpoints in late phase studies for drug development, these methods have clear utility in the area of clinical trials data analysis. A brief overview of methods for sample survey data analysis is first provided, followed by motivation for applying these methods to clinical trials data. Examples drawn from three clinical studies are provided to illustrate survey methods for logistic regression, proportional odds regression and proportional hazards regression. Potential problems with the proposed methods and ways of addressing them are discussed.

Sertraline versus imipramine to prevent relapse in chronic depression.

BACKGROUND: Chronic depressions are common, disabling and under-treated, and long-term treatment is little studied. We report the continuation phase results from a long-term treatment study. METHODS: After 12 weeks of acute phase treatment in a double-blind, randomized, parallel-group, multi-center trial of sertraline or imipramine, patients with chronic depression (> or = 2 years in major depression, or major depression superimposed on dysthymia) continued study drug for 16 weeks. Initially, 635 patients were randomized to sertraline or imipramine in a 2:1 ratio. Nonresponders after 12 weeks entered a 12-week double-blind crossover trial of the alternate medication. Entry into continuation treatment required at least a satisfactory response (partial remission) to initial or crossover treatment. RESULTS: Of 239 acute or crossover responders to sertraline, 60% entered continuation in full remission and 40% with a partial remission. These proportions were identical for imipramine patients (n = 147). For both drug groups, over two-thirds of those entering in full remission retained it. For those entering in partial remission, over 40% achieved full remission. Patients requiring crossover treatment were less likely to maintain or improve their response during continuation treatment. The two drugs did not differ significantly in response distribution, drop out rates or discontinuation due to side effects during continuation treatment. LIMITATIONS: The absence of a placebo group constrains interpretation of our results, but chronic depressions have low placebo response rates. CONCLUSIONS: Most chronic depression patients who remit with 12 weeks of sertraline or imipramine treatment maintain remission during 16 weeks of continuation treatment. Most patients with a satisfactory therapeutic response (partial remission) after 12 weeks of treatment maintain it or further improve. Patients treated with imipramine experienced more side effects, but both drugs were well tolerated.

Quality control of oncology clinical trials.

This article discusses issues that are essential to ensuring the reliability of the conclusions of oncology clinical trials. Though quality control is important at every stage of a well-run clinical trial, the authors focus on the quality of the data as evidenced by the results and conclusions of the study. Good quality control principles and practices are discussed for study planning, design, conduct, analysis, and interpretation.

Age of onset in chronic major depression: relation to demographic and clinical variables, family history, and treatment response.

BACKGROUND: The clinical and etiological significance of the early-late onset distinction in chronic major depressive disorder was explored. METHOD: Subjects were 289 outpatients with DSM-III-R chronic major depression drawn from a multi-site study comparing the efficacy of sertraline and imipramine in the acute and long-term treatment of chronic depression. Patients received comprehensive evaluations using semi-structured interviews and rating scales. RESULTS: Early-onset chronic major depression was associated with a longer index major depressive episode and higher rates of recurrent major depressive episodes, comorbid personality disorders, lifetime substance use disorders, depressive personality traits, and a history of psychiatric hospitalization. In addition, more early-onset patients tended to have a family history of mood disorders. The early-late onset distinction was not associated with differences in symptom severity, functional impairment, or treatment response. LIMITATIONS: Family members were not interviewed directly; there were a large number of statistical comparisons; and interrater reliability of the assessments was not evaluated. CONCLUSIONS: Early-onset chronic major depression has a more malignant course and is associated with greater comorbidity than late-onset chronic major depression.

Maintenance phase efficacy of sertraline for chronic depression: a randomized controlled trial.

CONTEXT: The chronic form of major depression is associated with a high rate of prevalence and disability, but no controlled research has examined the impact of long-term treatment on the course and burden of illness. OBJECTIVE: To determine if maintenance therapy with sertraline hydrochloride can effectively prevent recurrence of depression in the high-risk group of patients experiencing chronic major depression or major depression with antecedent dysthymic disorder ("double depression"). DESIGN: A 76-week randomized, double-blind, parallel-group study, conducted from September 1993 to November 1996. SETTING: Outpatient psychiatric clinics at 10 academic medical centers and 2 clinical research centers. INTERVENTION: Maintenance treatment with either sertraline hydrochloride (n = 77) in flexible doses up to 200 mg or placebo (n = 84). PATIENTS: A total of 161 outpatients with chronic major or double depression who responded to sertraline in a 12-week, double-blind, acute-phase treatment trial and continued to have a satisfactory therapeutic response during a subsequent 4-month continuation phase. MAIN OUTCOME MEASURE: Time to recurrence of major depression. RESULTS: Sertraline afforded significantly greater prophylaxis against recurrence than did placebo (5 [6%] of 77 in the sertraline group vs 19 [23%] of 84 in the placebo group; P = .002 for the log-rank test of time-to-recurrence distributions). Clinically significant depressive symptoms reemerged in 20 (26%) of 77 patients treated with sertraline vs 42 (50%) of 84 patients who received placebo (P = .001). With use of a Cox proportional hazards model, patients receiving placebo were 4.07 times more likely (95% CI, 1.51-10.95; P = .005) to experience a depression recurrence, after adjustment for study site, type of depression, and randomization strata. CONCLUSIONS: Maintenance therapy with sertraline is well tolerated and has significant efficacy in preventing recurrence or reemergence of depression in chronically depressed patients.

MSQ: Migraine-Specific Quality-of-Life Questionnaire. Further investigation of the factor structure.

MSQ, the 16-item Migraine-Specific Quality-of-Life Questionnaire (Version 1.0), was developed by Glaxo Wellcome Inc. to assess the effect of migraine and its treatment on patients' health-related quality of life (HR-QOL). The MSQ was hypothesised to measure 3 meaningful dimensions: (i) Role Function-Restrictive; (ii) Role Function-Preventive; and (iii) Emotional Function. The objective of this research was to further investigate the number of dimensions as well as the items contained in each dimension through principal components factor analysis of clinical trial data. Secondary objectives were to determine whether the factor structure changed in post-treatment visits compared with screening visits, to make recommendations for coding the MSQ when the patient did not have a migraine in the previous 4 weeks, and to modify the MSQ if so indicated by this research. Results supported the existence of 3 distinct factors which agreed strongly with the hypothesised dimensions. The analysis of post-treatment data suggested that the underlying factor structure of the MSQ varies as a result of treatment. Based on evaluations of the 'did not have a migraine' response, it was concluded that it be dropped from the MSQ. All these changes have been incorporated into MSQ (Version 2.0) which is being evaluated in studies to determine if its psychometric properties are different than the properties of the previous version.

Predictors of emotional numbing in posttraumatic stress disorder.

Little is known about the mechanisms underlying emotional numbing (EN). The functional relationship between other classes of posttraumatic stress disorder (PTSD) symptoms and EN is also not well understood. In the present study, we examined the statistical predictors of EN. We hypothesized that the severity of EN would be most strongly associated with the hyperarousal symptoms rather than the avoidance symptoms of PTSD, or comorbid depression or substance abuse. This prediction was derived from psychological and biological models that posit EN to be a product of the depletion of emotional resources subsequent to chronic hyperarousal. Using hierarchical multiple regression in two separate samples of Vietnam combat veterans, we found hyperarousal symptoms to be the most robust predictor of EN. These data suggest that there is a substantive relationship between hyperarousal symptoms and EN in PTSD.

Urinary cotinine and parent history (questionnaire) as indicators of passive smoking and predictors of lower respiratory illness in infants.

Studies of the effects of passive smoking on lower respiratory illness (LRI) have relied on questionnaires to measure exposure. We studied the association between two measures of passive smoking and the incidence of acute LRI in infants. We analyzed data from a community-based cohort study of respiratory illness during the first year of life in North Carolina. The incidence of LRI was determined by telephone calls at 2-week intervals. Environmental, demographic, and psychosocial risk factors for LRI were measured during home interviews. Tobacco smoke exposure was measured as the mean number of cigarettes smoked per day in the infant's presence. Smoke absorption by the infants was measured by the urinary cotinine/ creatinine ratio. Of the 485 infants in the study, 325 (67%) had telephone follow-up and at least two home interviews. In bivariate analyses, reported tobacco smoke exposure and urinary cotinine were associated with LRI. Only the association between reported exposure and LRI remained significant after adjusting for confounders, [adjusted incidence of LRI (episodes/child-year) non-exposed: 0.6; < or = 10 cigarettes/day: 0.9 (RR 1.5, 95% CI: 1.1, 2.0); > 10 cigarettes/day: 1.3 (RR 2.2, 95% CI: 1.3, 3.8)]. We conclude that infants reportedly exposed to tobacco smoke have an increased incidence of LRI. There are differences between questionnaire and biochemical measures of passive smoking. Urinary cotinine will not necessarily improve the validity of studies of the relationship of passive smoking to LRI in infants.

Predictors of Pneumocystis carinii pneumonia in HIV-infected persons. Pulmonary Complications of HIV Infection Study Group.

The Pulmonary Complications of HIV Infection Study is a prospective, multicenter, observational study evaluating pulmonary disease among HIV-infected persons. For approximately 52 mo, 1,182 HIV-infected subjects were followed. All participants were evaluated for pulmonary disease on a predetermined schedule. There were 145 episodes of Pneumocystis carinii pneumonia (PCP). Low CD4 count correlated with risk of PCP (p < 0.0001); 79% had CD4 counts less than 100/microl and 95% had CD4 counts less than 200/microl. Subtle changes in diffusing capacity for carbon monoxide (DLCO) were associated with PCP. Univariate analysis identified recurrent undiagnosed fevers, night sweats, oropharyngeal thrush, and unintentional weight loss to be associated with risk among persons with CD4 counts above 200/microl. Subjects in whom CD4 counts declined to below 200/microl and who were not receiving preventive therapy were nine times more likely to develop PCP within 6 mo compared with subjects who received such therapy. A strong trend toward differences between the sexes was detected. Black subjects had less than one third the risk of developing PCP as did white subjects (p < 0.0001). There was no significant difference in risk by HIV transmission category, study site, frequency of follow-up, age, education, smoking history, or use of antiretroviral therapy. Multivariable analysis revealed low CD4 lymphocyte count (p < 0.0001), use of prophylaxis (p < 0.0001), racial differences (p < 0.0001), and declining DLCO (p = 0.015) to influence risk. Constitutional signs and symptoms indicate increased risk for PCP among HIV-infected persons with CD4 counts above 200/microl.

Respiratory disease trends in the Pulmonary Complications of HIV Infection Study cohort. Pulmonary Complications of HIV Infection Study Group.

We examined trends in the incidence of specific respiratory disorders in a multicenter cohort with progressive human immunodeficiency virus (HIV) disease during a 5-yr period. Individuals with a wide range of HIV disease severity belonging to three transmission categories were evaluated at regular intervals and for episodic respiratory symptoms using standard diagnostic algorithms. Yearly incidence rates of respiratory diagnoses were assessed in the cohort as a whole and according to CD4 count or HIV transmission category. The most frequent respiratory disorders were upper respiratory tract infections, but the incidence of lower respiratory tract infections increased as CD4 counts declined. Specific lower respiratory infections followed distinctive patterns according to study-entry CD4 count and transmission category. Acute bronchitis was the predominant lower respiratory infection of cohort members with entry CD4 counts > or = 200 cells/mm3. In cohort members with entry CD4 counts of 200 to 499 cells/mm3, the incidence of bacterial and Pneumocystis carinii pneumonia each increased an average of 40% per year. In members with entry CD4 counts < 200 cells/mm3, acute bronchitis, bacterial pneumonia, and P. carinii pneumonia occurred at high rates without discernible time trends, despite chemoprophylaxis in more than 80% after Year 1, and the rate of other pulmonary opportunistic infections increased over time. Each year, injecting drug users had a higher incidence of bacterial pneumonia than did homosexual men. The yearly rate of tuberculosis was < 3 episodes/100 person-yr in each entry CD4 and HIV-transmission group. We conclude that the time trends of HIV-associated respiratory disorders are determined by HIV disease stage and influenced by transmission category. Whereas acute bronchitis is prevalent during all stages of HIV infection, incidence rates of bacterial pneumonia and P. carinii pneumonia rise continuously during progression to advanced disease. In advanced disease, the incidence of acute bronchitis, bacterial pneumonia and P. carinii pneumonia is high despite widespread chemoprophylaxis.

Innovative strategies using SUDAAN for analysis of health surveys with complex samples.

Large-scale health surveys provide a wealth of information for addressing problems in health sciences research. Designed for multiple purposes, these surveys frequently have large sample sizes and extensive measurements of demographic and socioeconomic characteristics, risk factors, disease outcomes and health care service use and costs. Complex features of the sampling design typically employed to select the survey sample, coupled with the vast amount of information available from the survey database, underlie issues that must be addressed during data processing and analysis. Numerous articles in the literature have focused on the debate of whether or not, and how, to control for features of the sample design during data analysis. Traditional statistical methods for simple random samples and the software that accompanies them have historically not had the capacity to account for the survey design. Recent advancements in statistical methodology for survey data analysis have greatly expanded the analytical tools available to the survey analyst. Commercial software packages that incorporate these methods offer the analyst convenient ways for applying such tools to large survey databases in an easy and efficient manner. We present an overview of analysis strategies for survey data and illustrate their application via the SUDAAN software system. Examples for analyses are provided through data from two large US health surveys, the National Health Interview Survey and the Longitudinal Study of Aging. Questions of both a cross-sectional and longitudinal nature are addressed. The examples involve logistic regression, time-to-event analysis, and repeated measures analysis.

Evaluation of a home-based intervention program to reduce infant passive smoking and lower respiratory illness.

We conducted a randomized controlled trial to determine whether a home-based intervention program could reduce infant passive smoking and lower respiratory illness. The intervention consisted of four nurse home visits during the first 6 months of life, designed to assist families to reduce the infant's exposure to tobacco smoke. Among the 121 infants of smoking mothers who completed the study, there was a significant difference in trend over the year between the intervention and the control groups in the amount of exposure to tobacco smoke; infants in the intervention group were exposed to 5.9 fewer cigarettes per day at 12 months. There was no group difference in infant urine cotinine excretion. The prevalence of persistent lower respiratory symptoms was lower among intervention-group infants of smoking mothers whose head of household had no education beyond high school: intervention group, 14.6%; and controls, 34.0%.

Application of sample survey methods for modelling ratios to incidence densities.

We describe ratio estimation methods for multivariately analysing incidence densities from prospective epidemiologic studies. Commonly used in survey data analysis, these ratio methods require minimal distributional assumptions and take into account the random variability in the at-risk periods. We illustrate their application with data from a study of lower respiratory illness (LRI) in children during the first year of life. One question of interest is whether children with passive exposure to tobacco smoke have a higher rate of LRI, on average, than those with no exposure and in a setting where age of child and season are taken into account. A second question is whether the relationship persists after adjusting for background variables such as family's socioeconomic status, crowding in the home, race, and type of feeding. The basic strategy consists of a two-step process in which we first estimate subgroup-specific incidence densities and their covariance matrix via a first-order Taylor series approximation. These estimates are used to test for differences in marginal rates of LRI between children exposed to tobacco smoke and those not exposed. We then fit a log-linear model to the estimated ratios in order to test for significant covariate effects. The ability to produce direct estimates of adjusted incidence density ratios for risk factors of interest is an important advantage of this approach. For comparison purposes and to address the limitations of the ratio method with respect to the number of covariates that can be controlled simultaneously, we consider survey logistic regression methods for the example data as well as logistic and Poisson regression models fitted via generalized estimating equation methods. Although the analysis strategy is illustrated with illness data from an epidemiologic study, the context of application is broader and includes, for example, data on adverse events from a clinical trial.

Lower respiratory illness in infants and low socioeconomic status.

OBJECTIVES: Infants from families of low socioeconomic status are said to suffer higher rates of lower respiratory illness, but this assertion has not been carefully examined. METHODS: We studied the frequency and determinants of lower respiratory illness in infants of different socioeconomic status (n = 393) by analyzing data from a community-based cohort study of respiratory illness during the first year of life in central North Carolina. RESULTS: The incidence of lower respiratory illness was 1.41 in the low socioeconomic group, 1.26 in the middle group, and 0.67 in the high group. The prevalence of persistent respiratory symptoms was 39% in infants in the low socioeconomic group, 24% in infants in the middle group, and 14% in infants in the high group. The odds of persistent respiratory symptoms in infants of low and middle socioeconomic status were reduced after controlling for environmental risk factors for lower respiratory illness. Enrollment in day care was associated with an increased risk of persistent symptoms among infants of high but not low socioeconomic status. CONCLUSIONS: Infants of low socioeconomic status are at increased risk of persistent respiratory symptoms. This risk can be partly attributed to environmental exposures, most of which could be changed.

Infant Health and Development Program for low birth weight, premature infants: program elements, family participation, and child intelligence.

The Infant Health and Development Program was an eight-site randomized controlled trial testing the efficacy of early intervention to enhance the cognitive, behavioral, and health status of low birth weight, premature infants. The 377 intervention families received for the first 3 years of life: (1) pediatric follow-up, (2) home visits, (3) parent support groups, and (4) a systematic educational program provided in specialized child development centers. The control group (n = 608) received the same pediatric follow-up and referral services only. This paper describes the delivery of the intervention and its outcomes. A Family Participation Index that was the sum of participation frequencies in each of the program modalities unique to the intervention revealed that program implementation was not different across the eight sites. Index scores did not vary systematically with mother's ethnicity, age, or education or with child's birth weight, gender, or neonatal health status; but they were positively related to children's IQ scores at age 3. Only 1.9% of children of families in the highest tercile of participation scored in the mentally retarded range (IQ less than or equal to 70), whereas 3.5% and 13% of children in the middle and lowest participation terciles, respectively, scored in the retarded range. Similar findings were obtained for borderline intellectual functioning. These findings are consistent with previous research linking intensity of intervention services with degree of positive cognitive outcomes for high-risk infants. The determinants of variations in individual family participation remain unknown.

An overview of statistical issues and methods of meta-analysis.

A meta-analysis is a statistical analysis of the data from some collection of studies in order to synthesize the results. In this paper we discuss issues that frequently arise in meta-analysis and give an overview of the methods used, with particular attention to the use of fixed- and random-effects approaches. The methods are then applied to two sample datasets.