Biomedical subjects
Publications and source records attributed to L J Wei.
In this note we consider the problem of drawing inference about the regression parameters in a linear model with survival data. A simple procedure based on the Buckley-James (1979, Biometrika 66, 429-436) estimating equation is proposed and illustrated with an example.
This paper considers clinical trials comparing two treatments with dichotomous responses where the data are examined periodically for early evidence of treatment difference. The existing group sequential methods for such trials are based on the large-sample normal approximation to the joint distribution of the estimators of treatment difference over interim analyses. We demonstrate through extensive numerical studies that, for small and even moderate-sized trials, these approximate procedures may lead to tests with supranominal size (mainly when unpooled estimators of variance are used) and confidence intervals with under-nominal coverage probability. We then study exact methods for group sequential testing, repeated interval estimation, and interval estimation following sequential testing. The new procedures can accommodate any treatment allocation rules. An example using real data is provided.
This note considers the problem of comparing the survival distributions of two treatment groups in a clinical trial where the treatment difference is measured in terms of a time-scale change. Patients enter the study sequentially and may be subject to random loss to follow-up. The survival data are reviewed periodically for early evidence of the treatment difference. An approach to constructing repeated confidence intervals for the scale-change parameter is described. These intervals provide useful information about the magnitude of the treatment difference in interim analyses. An example taken from an AIDS clinical trial is given.
Between 1975 and 1988, 92 pediatric patients have undergone bone marrow transplantation (BMT) at our institution for malignant or immune deficiency disease. We evaluated in a retrospective fashion 64 of these patients who survived beyond the first 60 days post-BMT. The clinical course was divided into: less than 60 days post-BMT (early) and greater than 60 days post-BMT (late). The presence or absence of renal insufficiency was noted as well as all known potential factors predisposing to insufficiency. Step-wise regression analysis was then performed to determine which of the factors were most significantly associated with renal dysfunction during the two periods. The follow-up period was 2 months to 11 years (mean 17.5 months). The mean age of the patients was 7.6 years (1 month-18 years). Fifty percent of the patients had renal insufficiency during the early period and 28% of the patients had insufficiency after the initial 60 days. Three major predictors of renal insufficiency were discovered. Cyclosporin A or amphotericin B early or late post-BMT was independently predictive of developing insufficiency during the same period. Conditioning with total body irradiation was a predictor for insufficiency in both periods. Early insufficiency was not predictive of late insufficiency. Hypertension was present in 31% of patients during the early period and in 16% during the late period. Hypertension was strongly associated with cyclosporin use and renal insufficiency. Renal insufficiency is a frequent sequela in children following BMT and likely results from a combination of radiation injury and drug toxicity.
Detection of elevated levels of immunoreactive trypsinogen (IRT) in dried neonatal blood spots has been used as a screening test for cystic fibrosis. In other cystic fibrosis newborn-screening studies, a sweat chloride test is generally performed only if an infant has a persistent IRT level above a selected cutoff value on both the initial and subsequent specimens. Neither the timing of the second specimen nor the value of the cutoff point for the second specimen has been comprehensively evaluated. In this randomized, controlled study, 145,024 infants were screened in the neonatal period for cystic fibrosis using the 99.8 percentile (180 ng/mL) as the neonatal cutoff point. A total of 129 infants had elevated neonatal IRT levels and had negative results on sweat tests (false-positive by IRT screening). A total of 54 children with cystic fibrosis were identified in the screened and comparison groups. Excluding patients with meconium ileus, 4 infants with cystic fibrosis had neonatal IRT values less than 180 ng/mL, and an additional 9 infants with cystic fibrosis had values decline to less than 180 ng/mL within the first 2 1/2 months of age. The IRT values of infants with and without cystic fibrosis overlapped considerably beyond 30 days of age. These findings suggest that further refinement of cystic fibrosis screening methodology will be necessary to achieve an acceptable sensitivity and specificity.
In this article we review the important statistical properties of the urn randomization (design) for assigning patients to treatment groups in a clinical trial. The urn design is the most widely studied member of the family of adaptive biased-coin designs. Such designs are a compromise between designs that yield perfect balance in treatment assignments and complete randomization which eliminates experimental bias. The urn design forces a small-sized trial to be balanced but approaches complete randomization as the size of the trial (n) increases. Thus, the urn design is not as vulnerable to experimental bias as are other restricted randomization procedures. In a clinical trial it may be difficult to postulate that the study subjects constitute a random sample from a well-defined homogeneous population. In this case, a randomization model provides a preferred basis for statistical inference. We describe the large-sample permutational null distributions of linear rank statistics for testing the equality of treatment groups based on the urn design. In general, these permutation tests may be different from those based on the population model, which is equivalent to assuming complete randomization. Poststratified subgroup analyses can also be performed on the basis of the urn design permutational distribution. This provides a basis for analyzing the subset of patients with observed responses when some patients' responses can be assumed to be missing-at-random. For multiple mutually exclusive strata, these tests are correlated. For this case, a combined covariate-adjusted test of treatment effect is described. Finally, we show how to generalize the urn design to a prospectively stratified trial with a fairly large number of strata.
The statistical properties of simple (complete) randomization, permuted-block (or simply blocked) randomization, and the urn adaptive biased-coin randomization are summarized. These procedures are contrasted to covariate adaptive procedures such as minimization and to response adaptive procedures such as the play-the-winner rule. General recommendations are offered regarding the use of complete, permuted-block, or urn randomization. In a large double-masked trial, any of these procedures may be acceptable. For a given trial, the relative merits of each procedure should be carefully weighed in relation to the characteristics of the trial. Important considerations are the size of the trial, overall as well as within the smallest subgroup to be employed in a subgroup-specific analysis, whether or not the trial is to be masked, and the resources needed to perform the proper randomization-based permutational analysis.
In comparing the effectiveness of two treatments, suppose that nondecreasing repeated measurements of the same characteristic are scheduled to be taken over a common set of time points for each study subject. A class of univariate one-sided global asymptotically distribution-free tests is proposed to test the equality of the two treatments. The test procedures allow different patterns of missing observations in the two groups to be compared, although the missing data mechanism is required to be independent of the observations in each treatment group. Test-based point and interval estimators of the global treatment difference are given. Multiple inference procedures are also provided to examine the time trend of treatment differences over the entire study. The proposed methods are illustrated by an example from a bladder cancer study.
We present methods for the analysis of a K-variate binary measure for two independent groups where some observations may be incomplete, as in the case of K repeated measures in a comparative trial. For the K 2 X 2 tables, let theta = (theta 1,..., theta K) be a vector of association parameters where theta k is a measure of association that is a continuous function of the probabilities pi ik in each group (i = 1, 2; k = 1,..., K), such as the log odds ratio or log relative risk. The asymptotic distribution of the estimates theta = (theta 1,..., theta K) is derived. Under the assumption that theta k = theta for all k, we describe the maximally efficient linear estimator theta of the common parameter theta. Tests of contrasts on the theta are presented which provide a test of homogeneity Ha: theta k = theta l for all k not equal to l. We then present maximally efficient tests of aggregate association Hb: theta = theta 0, where theta 0 is a given value. It is shown that the test of aggregate association Hb is asymptotically independent of the preliminary test of homogeneity Ha. These methods generalize the efficient estimators of Gart (1962, Biometrics 18, 601-610), and the Cochran (1954, Biometrics 10, 417-451), Mantel-Haenszel (1959, Journal of the National Cancer Institute 22, 719-748), and Radhakrishna (1965, Biometrics 21, 86-98) tests to nonindependent tables. The methods are illustrated with an analysis of repeated morphologic evaluations of liver biopsies obtained in the National Cooperative Gallstone Study.
Suppose that subjects are observed repeatedly over a common set of time points with possibly time-dependent covariates and possibly missing observations. At each time point we model the marginal distribution of the response variable and the effect of the covariates on that distribution using a class of quasi-likelihood models studied in McCullagh and Nelder. No parametric model of dependence of the repeated observations of the subject is assumed. For large samples, the quasi-likelihood estimates of the time-specific regression coefficients over the set of predetermined time points are shown to be approximately jointly normal. This, coupled with various inference procedures, provides a global picture about the effects of the covariates on the response variable over the entire study period. A lack-of-fit test for testing the adequacy of the assumed quasi-likelihood model is also provided. All the methods considered here are illustrated with real-life examples.
For the past two decades the Cox proportional hazards model has been used extensively to examine the covariate effects on the hazard function for the failure time variable. On the other hand, the accelerated failure time model, which simply regresses the logarithm of the survival time over the covariates, has seldom been utilized in the analysis of censored survival data. In this article, we review some newly developed linear regression methods for analysing failure time observations. These procedures have sound theoretical justification and can be implemented with an efficient numerical method. The accelerated failure time model has an intuitive physical interpretation and would be a useful alternative to the Cox model in survival analysis.