Frederick Mosteller

The comparative effects of postoperative analgesic therapies on pulmonary outcome: cumulative meta-analyses of randomized, controlled trials.

We performed meta-analyses of randomized, control trials to assess the effects of seven analgesic therapies on postoperative pulmonary function after a variety of procedures: epidural opioid, epidural local anesthetic, epidural opioid with local anesthetic, thoracic versus lumbar epidural opioid, intercostal nerve block, wound infiltration with local anesthetic, and intrapleural local anesthetic. Measures of forced expiratory volume in 1 s (FEV1), forced vital capacity (FVC), vital capacity (VC), peak expiratory flow rate (PEFR), PaO2, and incidence of atelectasis, pulmonary infection, and pulmonary complications overall were analyzed. Compared with systemic opioids, epidural opioids decreased the incidence of atelectasis (risk ratio [RR] 0.53, 95% confidence interval [CI] 0.33-0.85) and had a weak tendency to reduce the incidence of pulmonary infections (RR 0.53, 95% CI 0.18-1.53) and pulmonary complications overall (RR 0.51, 95% CI 0.20-1.33). Epidural local anesthetics increased PaO2 (difference 4.56 mm Hg, 95% CI 0.058-9.075) and decreased the incidence of pulmonary infections (RR 0.36, 95% CI 0.21-0.65) and pulmonary complications overall (RR 0.58, 95% CI 0.42-0.80) compared with systemic opioids. Intercostal nerve blockade tends to improve pulmonary outcome measures (incidence of atelectasis: RR 0.65, 95% CI 0.27-1.57, incidence of pulmonary complications overall: RR 0.47, 95% CI 0.18-1.22), but these differences did not achieve statistical significance. There were no clinically or statistically significant differences in the surrogate measures of pulmonary function (FEV1, FVC, and PEFR). These analyses support the utility of epidural analgesia for reducing postoperative pulmonary morbidity but do not support the use of surrogate measures of pulmonary outcome as predictors or determinants of pulmonary morbidity in postoperative patients. Implications: When individual trials are unable to produce significant results, it is often because of insufficient patient numbers. It may be impossible for a single institution to study enough patients. Meta-analysis is a useful tool for combining the data from multiple trials to increase the patient numbers. These meta-analyses confirm that postoperative epidural pain control can significantly decrease the incidence of pulmonary morbidity. (Anesth Analg 1998;86:598-612)

Some Statistical Methods for Combining Experimental Results

Advances in science and technology are generally the product of multiple investigations. This article discusses statistical methods for combining empirical results from a series of different experiments or clinical investigations. We delineate the steps an assessor might take in combining data from different studies and provide references for topics not discussed in detail. The article reviews some of the most commonly used statistical techniques for combining results in the medical and social sciences.

Reporting on Methods in Clinical Trials

Abstract A clinical trial cannot be adequately interpreted without information about the methods used in the design of the study and the analysis of the results. To determine the frequency of reporting what we consider 11 important aspects of design and analysis, we surveyed all 67 clinical trials published in the New England Journal of Medicine, the Lancet, and the British Medical Journal from July through December 1979 and in the Journal of the American Medical Association from July 1979 through June 1980. Of all 11 items in the 67 trials published in all four journals, 56 per cent were clearly reported, 10 per cent were ambiguously mentioned, and 34 per cent were not reported at all. At least 80 per cent of the 67 trials reported information about statistical analyses, statistical methods used, and random allocation of subjects, yet only 19 per cent reported the method of randomization. Loss to follow-up was discussed in 79 per cent of the articles, treatment complications in 64 per cent, and admission...

Recognizing the Maximum of a Sequence

Abstract The classical dowry, secretary, or beauty contest problem is extended in several directions. In trying to find sequentially the maximum of a random sequence of fixed length, the chooser can have one or several choices (section 2), no information about the distribution of the values (section 2), or at the other extreme, full information about the distribution and the value of the observation itself (section 3). He can have an opponent who alters the properties of the sequence (section 4). The payoff function may be 0 or 1 (sections 2–4), or it may be the value of the observation itself as in certain investment problems (section 5). Both theoretical and numerical results are given for optimum and near optimum play.

Association and Estimation in Contingency Tables

OUR 1967 Committee on Publications, chaired by David L. Wallace, found that many American Statistical Association members desired more review and survey papers. These have been hard to come by, and so as my last act before leaving office, I decided to provide a short survey paper on some related ideas in a field where nearly all of us sometimes work-that of contingency tables. These ideas are largely available in the literature, and yet they have not often been put together, though I. J. Goods monograph [9] and Leo Goodmans many papers form good sources. But my paper is not intended as a review of the literature, only as a survey of one set of ideas about estimation in the analysis of contingency tables. I fear that the first act of most social scientists upon seeing a contingency table is to compute chi-square for it. Sometimes this process is enlightening, sometimes wasteful, but sometimes it does not go quite far enough.

A Mathematical Model for Simple Learning

A mathematical model for simple learning is presented. Changes in the probability of occurrence of a response in a small time h are described with the aid of mathematical operators. The parameters which appear in the operator equations are related to experimental variables such as the amount of reward and work. Relations between the probability and empirical measures of rate of responding and latent time are defined. Acquisition and extinction of behavior habits are discussed for the simple runway and for the Skinner box. Equations of mean latent time as a function of trial number are derived for the runway problem; equations for the mean rate of responding and cumulative numbers of responses versus time are derived for the Skinner box experiments. An attempt is made to analyze the learning process with various schedules of partial reinforcement in the Skinner type experiment. Wherever possible, the correspondence between the present model and the work of Estes [2] is pointed out.

Postoperative patient-controlled analgesia: Meta-analyses of initial randomized control trials

STUDY OBJECTIVE To compare outcomes during conventional analgesia (as-needed intramuscular dosing) and patient-controlled analgesia (PCA) in postoperative patients by analyzing data from published comparative trials. DESIGN Meta-analyses of 15 randomized control trials. PATIENTS Seven hundred eighty-seven adult patients (aged 16 to 65) undergoing various operative procedures. INTERVENTIONS Either PCA or conventional analgesia for postoperative pain control. MEASUREMENTS AND MAIN RESULTS Data were extracted on analgesic efficacy, analgesic use, patient satisfaction, length of hospital stay, and side effects. Meta-analyses of the data showed the following: (1) greater analgesic efficacy when PCA was used, with a mean additional benefit of 5.6 on a scale of 0 to 100 (SED, 2.2; p = 0.006); (2) a nonsignificant trend toward reduced analgesic use in PCA patients, based on a count of trials finding in one direction or the other (p = 0.092); (3) a 42% difference in the proportion of patients expressing satisfaction over dissatisfaction (SED, 20%; p = 0.02), with PCA being preferred; (4) a nonsignificant trend toward shortening of length of hospital stay with PCA use (mean, 0.15 days, SED, 0.13; p = 0.24); (5) no significant differences in the occurrence of any side effect. CONCLUSIONS Patient preference strongly favors PCA over conventional analgesia. Patients using PCA also obtain better pain relief than those using conventional analgesia, without an increase in side effects. Favorable effect of PCA upon analgesic usage and length of hospital stay did not in the initial trials attain statistical significance. Nonetheless, the favorable trends in the mean effect sizes for both outcomes argue that further studies of both outcomes should be performed to determine whether the favorable impact of PCA upon either may become statistically significant if larger numbers of patients are enrolled.

How study design affects outcomes in comparisons of therapy. II: Surgical

We analysed the results of 221 comparisons of an innovation with a standard treatment in surgery published in six leading surgery journals in 1983 to relate features of study design to the magnitude of gain. For each comparison we measured the gain attributed to the innovation over the standard therapy by the Mann-Whitney statistic and the difference in proportion of treatment successes. For primary treatments (aimed at curing or ameliorating a patients principal disease), an average gain of 0.56 was produced by 20 randomized controlled trials. This was less than the 0.62 average for four non-randomized controlled trials, 0.63 for 19 externally controlled trials, and 0.57 for 73 record reviews (0.50 represents a toss-up between innovation and standard). For secondary therapies (used to prevent or treat complications of therapy), the average gain was 0.53 for 61 randomized controlled trials, 0.58 for eleven non-randomized controlled trials, 0.54 for eight externally controlled trials, and 0.55 for 18 record reviews. Readers of studies evaluating new treatments, particularly for primary treatments, may consider adjustment of the gain according to the study type.

Guidelines for statistical reporting in articles for medical journals. Amplifications and explanations.

The 1988 edition of the Uniform Requirements for Manuscripts Submitted to Biomedical Journals includes guidelines for presenting statistical aspects of scientific research. The guidelines are intended to aid authors in reporting the statistical aspects of their work in ways that are clear and helpful to readers. We examine these guidelines for statistics using 15 numbered statements. Although the information presented relates to manuscript preparation, it will also help investigators in earlier stages make critical decisions about research approaches and protocols.

On Some Useful “Inefficient” Statistics

Several statistical techniques are proposed for economically analyzing large masses of data by means of punched-card equipment; most of these techniques require only a counting sorter. The methods proposed are designed especially for situations where data are inexpensive compared to the cost of analysis by means of statistically “efficient” or “most powerful” procedures. The principal technique is the use of functions of order statistics, which we call systematic statistics.