David P. Byar

Randomized clinical trials. Perspectives on some recent ideas.

In spite of the controversy over the role of randomized clinical trials in medical research, the rationale underlying such trials remains persuasive as compared to recent suggestions for alternative non-randomized studies such as those relying on the use of historical controls and adjustment technics. Others have suggested that recent statistical innovations for improving clinical trials, including adaptive allocation of treatment to patients and sequential stopping procedures, are underutilized. These innovations, though theoretically interesting, are not easily adapted to large-scale, complex medical trials in which there may be multiple end points and delayed response times. Ethical considerations suggest that randomized trials are more suitable than uncontrolled experimentation in protecting the interests of patients. Randomized clinical trials remain the most reliable method for evaluating the efficacy of therapies.

Why Data Bases Should Not Replace Randomized Clinical Trials

Advances in computer technology have made it possible to store large amounts of observational data concerning treatment of patients for medical disorders. It has been suggested that these data banks might replace randomized clinical trials as a means of evaluating the efficacy of therapies. A review of the methodological problems likely to arise in analyzing such data for the purpose of comparing treatments suggests that sound inferences would not generally be possible because of difficulties with bias in treatment assignment, nonstandard definitions, definitions changing in time, specification of groups to be compared, missing data, and multiple comparisons.

COMPARISONS OF PLACEBO, PYRIDOXINE, AND TOPICAL THIOTEPA IN PREVENTING RECURRENCE OF STAGE I BLADDER CANCER*

Animal studies have shown that metabolites of tryptophan can cause bladder cancer, and human observations reveal an appreciable incidence of abnormalities of tryptophan metabolism in patients with bladder cancer. It has been suggested that pyridoxine (vitamin B6) may correct this abnormality and prevent recurrences of superficial bladder cancers. Intravesical instillation of thiotepa has been used for more than fifteen years in the treatment of superficial bladder cancer, but no controlled trials have been done. We report here a prospective clinical trial of 121 patients with Stage I bladder cancer randomized to placebo, pyridoxine, or intravesical thiotepa. The percentages of patients with recurrences over the period of study were 60.4, 46.9, and 47.4 for the three groups, respectively, and did not differ significantly. However, if patients having recurrences during the first ten months or followed up less than ten months were excluded, pyridoxine was significantly better than placebo (P = 0.03). Thiotepa significantly reduced the recurrence rate compared with placebo (P = 0.016) or pyridoxine (P = 0.015). These results suggest that a new trial of pyridoxine should be undertaken in which the tryptophan metabolites are measured and that further study of intravesical instillation of chemotherapeutic agents is warranted.

Incidence of cardiovascular disease and death in patients receiving diethylstilbestrol for carcinoma of the prostate

Patients treated with a 5.0‐mg daily dose of diethylstilbestrol (DES) had an increased incidence of fatal and non‐fatal cardiovascular disease when compared to placebo in all stages of prostatic cancer (p < 0.025). The pretreatment cardiovascular status of estrogen‐treated patients was generally better than those treated with placebo. Therapy with DES 5.0 mg did not increase survival of Stage III or IV patients significantly when compared to placebo. The decrease in cancer mortality associated with the 5.0‐mg dose of DES was offset by an increase in deaths from cardiovascular causes. Early endocrine treatment of patients with asymptomatic Stage III carcinoma is not indicated. Endocrine therapy should be started early only in Stage IV patients. When DES is preferred, it should be administered in a dose lower than 5.0 mg. Complications of estrogen therapy may be due to an increased incidence of thromboembolism.

Aspects of statistical design for the community intervention trial for smoking cessation (COMMIT)

We present statistical considerations for the design of the Community Intervention Trial for Smoking Cessation (COMMIT). One outcome measurement, the quit rate in randomly selected cohorts of smokers, is compared with another outcome measurement, the decrease in smoking prevalence, in terms of statistical efficiency and interpretability. The COMMIT study uses both types of outcome measurements. The merits of pair-matching the communities are considered, and sample size calculations take into account heterogeneity among pair-matched communities. In addition to significance tests based on the permutational (randomization) distribution, we also describe approaches for covariate adjustment. The COMMIT design includes 11 pair-matched communities, which should provide good power to detect a 10% or greater difference in quit rates between the intervention and control communities in cohorts of heavy smokers and in cohorts of light or moderate smokers. The power is only moderate to detect intervention effects on the decreases in overall smoking prevalence or in the prevalence of heavy smoking.

How many controls

A COMMON question in clinical and epidemiologic research is, ‘How many controls are needed for this study ?. Some retrospective studies can be strengthened by using more controls than cases, and some prospective clinical studies can be improved by unequal allocation of subjects into treatment and control groups. We shall confine our discussion of ‘How many controls? to studies comparing responses in only two groups. The first group (G,) may be subjected to a new treatment (T,) while the second (G,) is given conventional or control therapy (T2), and the response may be quantitative, such as the lowering of blood pressure, or qualitative, such as whether or not the subject survives. In retrospective studies G1 is typically a group of ‘cases’ and Gz a group of matched or unmatched controls, and the response is whether or not each subject was exposed to a possible etiologic agent. Most such studies allocate equal numbers of subjects to G1 and G2, namely n, = n2 = n, and formulae, graphs, and tables are available for determining the sample size n, = n2 required to have a given probability (power) of detecting a prespecified treatment effect at significance level CI when equal allocation is used. Pertinent references include [l, pp. 15-311, [2, p. 191, [3], [4, pp. 4794821 and [S, pp. 111-114 and 221-2221. We now discuss three situations in which unequal allocation may be preferred. In prospective clinical trials involving treatments with comparable risk, the investigator may have discretion over the numbers treated in each group. Then other factors, such as the relative monetary costs of the two treatments, or the relative inconvenience or discomfort to patients, can influence the experimental design. If it costs r times as much to study a subject in G1 as in G2, and if one wishes to minimize the cost of the experiment while maintaining the same power (the power is 1 p, where fl is the probability of type II error as in [6, p. 279]), one should allocate Jr subjects to G2 for each subject allocated to G,. This result, which we call the ‘square root rule’, has been discussed by Cochran [7, p. 1453 and Nam [8]. We generalize the square root rule to the case where the response variable has different variances in each group. This generalized square

Percentage Points of the Asymptotic Distributions of One and Two Sample K-S Statistics for Truncated or Censored Data

In this article we provide tables of percentage points of the asymptotic distribution of the one sample truncated Kolmogorov-Smirnov statistic. We discuss use of the tables in goodness of fit problems involving tnmcated or censored data and indicate that the tables provide accurate critical values for sample sizes greater than 30. We also discuss use of the tables in situations involving censored data and in two sample testing problems.

Orchiectomy for advanced prostatic carcinoma A reevaluation

Abstract The efficacy of four treatments, placebo, orchiectomy plus placebo, estrogen (diethylstilbestrol 5.0 mg. daily), and orchiectomy plus estrogen, was evaluated in 1,903 Stage III and IV prostatic cancer patients by comparing (1) survival curves, (2) causes of death, (3) clinical response, (4) development of metastases in Stage III patients, and (5) incidence of treatment change. Survival curves did not differ significantly for the four treatment groups except in Stage III, in which orchiectomy plus estrogen was worse than placebo, or orchiectomy plus placebo. There were significantly more cancer deaths in the two nonestrogen groups in both stages. Otherwise, there were no appreciable differences between the three hormonal treatments. We conclude that estrogen is more effective than orchiectomy in preventing cancer deaths, and the addition of orchiectomy to estrogen does not offer a clear-cut advantage over either treatment alone. Therefore, if treatment becomes necessary because of cancer symptoms, initial treatment with estrogen is preferred.

Selecting optimal treatment in clinical trials using covariate information

Abstract We are interested in the question ‘which treatment is best for which kinds of patients?’ rather than the classical question, ‘which treatment is best overall?’. A survivorship function in which the hazard is a function of the covariates is fitted both for all treatments combined and for separate treatments. Likelihood ratio tests are used to detect significant treatment-covariate interactions. If there are none, we test for a best overall treatment. Otherwise we define an optimal treatment for each patient. By studying the combinations of covariates which lead to selection of the various treatments as optimal, we make recommendations of treatment for different kinds of patients.

Randomized Clinical Trials

In spite of the controversy over the role of randomized clinical trials in medical research, the rationale underlying such trials remains persuasive as compared to recent suggestions for alternative non-randomized studies such as those relying on the use of historical controls and adjustment technics. Others have suggested that recent statistical innovations for improving clinical trials, including adaptive allocation of treatment to patients and sequential stopping procedures, are underutilized. These innovations, though theoretically interesting, are not easily adapted to large-scale, complex medical trials in which there may be multiple end points and delayed response times. Ethical considerations suggest that randomized trials are more suitable than uncontrolled experimentation in protecting the interests of patients. Randomized clinical trials remain the most reliable method for evaluating the efficacy of therapies.