Sample size determination for testing whether an identified treatment is best.

Laska and Meisner (1989, Biometrics 45, 1139-1151) dealt with the problem of testing whether an identified treatment belonging to a set of k + 1 treatments is better than each of the other k treatments. They calculated sample size tables for k = 2 when using multiple t-tests or Wilcoxon-Mann-Whitney tests, both under normality assumptions. In this paper, we provide sample size formulas as well as tables for sample size determination for k > or = 2 when t-tests under normality or Wilcoxon-Mann-Whitney tests under general distribution assumptions are used.

Some new multiple-test procedures for dose finding.

In Tamhane, Hochberg, and Dunnett (1) we focused primarily on step-down test procedures based on contrasts among the sample means to find the minimum effective dose in a dose-response study. In the present article we use the global tests of Bartholomew (2,3) and Hayter (4) in these step-down procedures. We also propose a new step-down procedure that permits tests based on a class of contrasts [step and basin contrasts of Ruberg (5) are examples of such contrasts] that could not be used with the step-down procedures studied in our previous paper because of lack of control of the familywise error rate. A simulation study to compare the four procedures proposed in the present paper with the top four procedures from the previous article is carried out. It is found that the step-down procedure based on Bartholomew's test and the new step-down procedure based on step and modified basin contrasts generally perform better than the other procedures for a wide range of dose-response profiles.

Multiple testing to establish superiority/equivalence of a new treatment compared with kappa standard treatments.

In this paper we develop multiple hypotheses testing procedures to compare a new treatment with a set of standard treatments in a clinical trial. The aim is to classify the new treatment with respect to each of the standards, by specifying those to which the new treatment is superior, those to which the new treatment is equivalent and those to which one can establish neither superiority nor equivalence. We propose several stepwise procedures and compare them with respect to their familywise error rates and power. The step-down methods SD1 and SD2 test for superiority first, followed by tests for equivalence for those comparisons where we cannot establish superiority. The step-up methods SU1 and SU2 test for equivalence first, followed by tests for superiority for those comparisons where we can establish at least equivalence. The methods SD3 and SU3 apply the tests for superiority and equivalence in pairs. All the methods require that we specify a threshold value delta > 0 in advance for defining equivalence. In applications where it is not possible to specify a value delta, we can use the method SD1 by testing for superiority first, followed by one-sided confidence limits on the efficacy differences for those comparisons where we cannot establish superiority.

An alternative to the use of two-sided tests in clinical trials.

There is a controversy in the literature concerning the use of one- and two-sided tests in clinical trials. Some contend that, when the research question relates to improved efficacy or safety, that is, the expected change is in one direction only, the hypothesis test should reflect this by being one-sided. Others insist on the use of a two-sided test in case a treatment difference in the opposite direction to that expected might turn up. We propose an alternative procedure to the two-sided test which also provides protection against overlooking a negative effect. The proposed procedure tests simultaneously for a positive difference and for equivalence. We illustrate the procedure by applying it to the results of a recent clinical trial.

Multiple test procedures for dose finding.

The problem of identifying the lowest dose level for which the mean response differs from that at the zero dose level is considered. A general framework for stepwise testing procedures that use contrasts among the dose level means is proposed. Using this framework, several new procedures are derived. These and some existing procedures, including that of Williams (1971, Biometrics 27, 103-117; 1972, Biometrics 28, 519-531), are compared analytically and by an extensive simulation study for the normal theory balanced one-way layout case. It is pointed out that the procedures based on the so-called step and basin contrasts proposed by Ruberg (1989, Journal of American Statistical Association 84, 816-822) have excessively high type I familywise error rates (FWEs) and, hence, they should not be used. Some findings of the simulation study are as follows: For monotone dose mean configurations, Williams' procedure and two step-down test procedures based on Helmert and linear contrasts offer the best performance. For nonmonotone dose mean configurations, the performance of Williams' procedure does degrade somewhat, but the other two procedures are still the best. For more complex designs, a simple step-down test procedure that uses any alpha-level tests (not necessarily t-tests) to compare each dose level with the zero dose level controls the FWE and is the only alternative available, but its power is rather low, especially under nonmonotone configurations. Step-up procedures are generally dominated by step-down procedures when the same contrasts are used although the differences are not great.

Step-up multiple testing of parameters with unequally correlated estimates.

We consider the problem of simultaneously testing k > or = to 2 hypotheses on parameters theta(1), ..., theta(k) using test statistics t(1), ..., t(k) such that a specified familywise error rate alpha is achieved. Dunnett and Tamhane (1992a) proposed a step-up multiple test procedure, in which testing starts with the hypothesis corresponding to the least significant test statistic and proceeds towards the most significant, stopping the first time a significant test result is obtained (and rejecting the hypotheses corresponding to that and any remaining test statistics). The parameter estimates used in the t statistics were assumed to be normally distributed with a common variance, which was a known multiple of an unknown sigma(2), and known correlations which were equal. In the present article, we show how the procedure can be extended to include unequally correlated parameter estimates. Unequal correlations occur, for example, in experiments involving comparisons among treatment groups with unequal sample sizes. We also compare the step-up and step-down multiple testing approaches and discuss applications to some biopharmaceutical testing problems.

Comparisons between a new drug and active and placebo controls in an efficacy clinical trial.

D'Agostino and Heeren (DH) discussed the multiple comparison issues that arise in evaluating both sensitivity and efficacy in over-the-counter drug trials. We propose a general definition of sensitivity that includes DH's definition as a special case. We also propose a test for sensitivity that coincides with the MIN test of Laska and Meisner at one extreme but has the advantage of identifying specific drugs satisfying the sensitivity criterion when some fail to do so. We suggest that the test of Dunnett as well as an extension of it may be useful for the efficacy comparisons.

Step-down multiple tests for comparing treatments with a control in unbalanced one-way layouts.

We show how a well-known multiple step-down significance testing procedure for comparing treatments with a control in balanced one-way layouts can be applied in unbalanced layouts (unequal sample sizes for the treatments). The method we describe has the advantage that it provides p-values, for each treatment versus control comparison, that take account of the multiple step-down testing nature of the procedure. These joint p-values can be used with any value of alpha, the fixed type I family wise error rate bound, that may be specified by the investigator. To determine the p-values, it is necessary to compute a multivariate Student t integral, for which a computer program is available. This procedure is more powerful than the step-down Bonferroni procedure of Holm and the single-step procedure of Dunnett. An example from the pharmaceutical literature is used to illustrate the procedure.

Simultaneous interval estimates of the odds ratio in studies with two or more comparisons.

More than one odds ratio estimate will often arise from a single epidemiologic study. Examples of designs where this may occur include those where there is more than one case or control group, and investigations of several risk factors as part of the same study. Various methods for presenting multiple interval estimates are discussed, including: the naive method, the Bonferroni method, the Dunn method, the Scheffé method, and the Dunnett method. For rectangular regions the Dunnett method gives a region with the most appropriate confidence level, but this region contains a different set of odds ratio estimates than are implied by the usual significance tests. A confidence ellipse circumscribed by the Scheffé limits gives the best agreement with the significance tests. Each of these methods is illustrated with a numerical example.

A randomized trial of quality assurance in nursing homes.

Sixty nursing homes were randomly allocated to receive or not to receive a quality assurance intervention. The experimental intervention included the use of predeveloped quality assurance packages, the services of a quality assurance consultant, and the process of working through the quality assurance cycle with one of two principal indicator conditions. Two prevalent health problems, hazardous mobility and constipation, were selected as the principal indicator conditions. To detect co-intervention, one of two hidden secondary indicator conditions (potential skin breakdown and urinary incontinence) was assessed in each facility. In the control nursing homes, both the principal and secondary indicator conditions were hidden from staff. The care for 1,525 residents was examined before and after the intervention using a retrospective record review initiated for the study purposes. Improvement in management of the principal conditions, hazardous mobility and constipation, was greater in the experimental group (P less than 0.03 and P less than 0.005, respectively). Neither group changed its management of the hidden conditions. Behavior change was achieved using quality assurance-linked interventions. Further research should focus on refining quality assurance interventions that provide staff education and motivational strategies.

Lipid composition of rabbit aortic wall following removal of endothelium by balloon catheter.

To determine the lipid content of the neointima formed following balloon catheter deendothelialization, we measured the chemical composition of intimal medial samples of the aortic wall of rabbits at 6, 12, and 24 months after the operation. The areas of endothelial regrowth and the denuded areas were outlined by intravenous injection of Evan's blue dye. Lipids were extracted by chloroform/methanol (2:1, vol/vol) separated by thin-layer chromatography, eluted, and estimated by gas-liquid chromatography. No tendency to regression was observed. By contrast we observed a progressive increase over time in the concentration of most lipid classes in the areas of endothelial regrowth. There is evidence indicating that the increase in the lipid concentration parallels the increase in the glycosaminoglycan content in the areas of endothelial regrowth. It is possible that endothelial injury causes some irreversible changes in the composition of the neointima in areas of endothelial regeneration.

Correlation of elevated C1q binding activity and carcinoembryonic antigen levels with clinical features and prognosis in bronchogenic carcinoma.

The presence of immune complexes and carcinoembryonic antigen (CEA) was investigated in 50 patients with bronchogenic carcinoma at the time of and/or following diagnostic or definitive surgery. Immune complexes were measured by the C1q binding test and CEA by the Z gel method and elevations defined as values in excess of 2 S.D. above the normal mean, greater than or equal to 9.2% for C1q binding activity (C1q-BA) and greater than or equal to 5.0 ng/ml for CEA. The overall incidence of elevated values was 30.7% for C1q-BA and 34.2% for CEA. There was a greater incidence of elevated values of C1q-BA among patients with clinically evident disease. The differences with respect to CEA elevation were not significant due to the fact that 6 of 9 samples with elevated CEA values obtained from patients with no evident disease were in fact associated with the presence of clinically undetectable disease in these patients. Elevation of C1q-BA and CEA beyond the immediate postoperative period was predictive of a significantly shorter median survival time. The most significant differences in survival time were seen between patients with normal values for C1q-BA and CEA and those with elevations of one or both parameters, 6.0 vs. 19.5 months (p less than 0.001). Elevation of either parameter during the immediate pre- and postoperative period was not predictive of a poor survival. In terms of clinical application, it appears that CEA estimation had the best predictive value but that the addition of C1q-BA measurement may provide additional prognostic information, particularly in patients who do not have elevated CEA values.

How many well baby visits? A randomized trial in progress.

The objective of this study in progress is to determine whether decreasing the number of well-baby visits during the first two years of life from the currently allowable ten visits to five, can be done with efficacious and safe results. Five hundred babies under the care of their family physicians were randomly assigned to an experimental group receiving five well-baby visits in the first two years of life or to a control group receiving ten visits. This article presents the results pertaining to the collection of the cohort.

Application of tumor marker analysis to patients with lung cancer.

Measurement of Clq-BA and CEA levels in patients with lung cancer may provide additional information about their clinical status discriminating between disease free patients and those with residual disease. Post-operative determination of Clq-BA and CEA levels may assist in defining patients which have poor prognosis. Preliminary evidence suggests that Clq-BA measurement may provide additional prognostic information in a small group of patients with normal CEA values. Finally, the presence of elevated Clq-BA in tumor bearing patients is of fundamental importance in the biology of cancer as it suggests that there are some tumor associated antigens which provoke the production of antibody in the host with the resultant formation of circulating immune complexes. The definition of the nature of the antigen in these immune complexes awaits further study.

Self-recording of blood pressure in the management of hypertension.

The efficacy of self-recording of blood pressure in the management of hypertension was assessed in a randomized clinical trial involving 140 persons who had been receiving antihypertensive therapy for a year or more, but whose diastolic blood pressure had remained at 95 mm Hg or higher. To control for the increased attention implicit in self-recording, which might affect blood pressure, the patients were assigned at random to one of the four groups: self-recording and monthly home visits, self-recording only, monthly home visits only, and neither self-recording nor monthly home visits. This design also permitted assessment of the effect of home visits. During the 6-month experiment no significant differences were apparent between the groups in either compliance or diastolic blood pressure. However, both self-recording and monthly home visits produced a reduction in blood pressure among patients who admitted to difficulty remembering to take their pills; a reduction was not seen among patients who said they had no such difficulty. This confirmed an earlier observation suggesting that this easily identified group of patients may be the most responsive to intervention programs.

Measurement of carcinoembryonic antigen in patients with bronchogenic carcinoma.

Estimation of CEA levels by the Z-gel method indicates that smokers, patients with limited lung cancer and patients with extensive lung cancer have higher values than nonsmoking controls. The CEA levels within each group are significantly different from one another. Use of CEA estimation for diagnostic purposes is limited because of the considerable overlap between normal controls and patients with cancer, the relatively low incidence of elevated values in patients with limited disease and the high incidence of false negatives (20%) even in patients with extensive disease. Elevated CEA values are associated with a poor prognosis and could be of clinical value as an addition to clinical staging to determine survival particularly for patients with extra-thoracic disease. Persistently high values in patients deemed clinically disease-free postoperatively are indicative of residual disease and a poor prognosis. If and when effective therapy for bronchogenic carcinoma becomes available, monitoring of CEA values may be useful in some patients as an early indication of release. Further studies are required to determine if the extraordinarily poor prognosis associated with marked elevations of CEA may be used as an additional criterion of inoperability in such patients.

Significance testing to establish equivalence between treatments, with special reference to data in the form of 2X2 tables.

In this paper, the role of significance testing in establishing equivalence between treatments is described. In place of the more customary null hypothesis of no difference between treatments, the hypothesis that the true difference is equal to a specified delta is tested. The particular case of comparing two binomial samples is described in detail. Results using a method due to Gart and approximations based on the chi2 and normal distributions are compared. It is found that the test based upon chi2 with continuity correction agrees best overall with Gart's (1971) test and is therefore recommended for use when computational facilities needed for carrying out the latter are not available.

Efficacy of loxapine in the treatment of paranoid schizophrenia.

A combined analysis of data from 11 controlled studies of loxapine versus either chlorpromazine or trifluoperazine in acute schizophrenia (5 studies) and chronic schizophrenia (6 studies) showed statistically significant superiority of loxapine on several items and factors of standardized psychiatric rating scales. Upon review of these findings, it was observed that the rating scale symptoms for which loxapine appeared superior to the reference compounds could, in the main, be considered a broad paranoid "cluster". The data were then reanalyzed to detect possible differences in efficacy of loxapine versus the reference compounds in those patients with a clinical diagnosis of schizophrenia, paranoid type and in those patients of any diagnostic subtype other than paranoid. Results of these analyses demonstrated clear superiority of loxapine in paranoid schizophrenic patients; nonparanoid patients responded at least equally as well to loxapine as to the reference compounds. Findings could not be attributed to inadequate dosages of the references compounds or inequality of treatment groups at baseline.