Virginia F. Flack

Frequency of Selecting Noise Variables in Subset Regression Analysis: A Simulation Study

Abstract This article presents the results of a simulation study of variable selection in a multiple regression context that evaluates the frequency of selecting noise variables and the bias of the adjusted R 2 of the selected variables when some of the candidate variables are authentic. It is demonstrated that for most samples a large percentage of the selected variables is noise, particularly when the number of candidate variables is large relative to the number of observations. The adjusted R 2 of the selected variables is highly inflated.

Rehabilitation needs and breast cancer: The first month after primary therapy

SummaryBreast cancer is the most common neoplasm in occidental women, yet very little information is available about the rehabilitation needs of these patients. This report reviews in detail the physical, psychological, social, and financial problems identified by fifty women during the first month after primary surgical treatment. The most frequently reported problems were physical and psychological. Patients undergoing modified radical mastectomy (MRM) are compared with patients receiving segmental mastectomy and primary radiotherapy (SM). There were few differences between the two surgical treatment groups; however, patients in the MRM group reported more difficulty with clothing and body image, while the SM group had more disruption of recreation and social activities. Information about the rehabilitation problems of patients with this common neoplasm should be helpful to patients and to the members of the health care team who are helping them recover.

Radiographic Secondary Caries Prevalence in Teeth with Clinically Defective Restorations

Decisions to replace existing restorations are often based on clinical findings of margin discrepancies and other restoration defects. Previous studies have suggested that such findings do not correlate well with the actual presence of secondary caries, and that treatment should be deferred until caries is clinically or radiographically evident. The purpose of this study was to assess the frequency with which clinically defective restorations are associated with radiographic secondary caries. As part of a study to assess the efficacy of guidelines for the ordering of dental radiographs, 6285 restored teeth in 490 subjects were examined clinically and radiographically by three calibrated investigators. Specific criteria were used to determine whether restorations were intact or defective. Of the total, 822 teeth (13%) were judged to have clinically defective restorations. Of these, 113 teeth (14%) had radiographic secondary caries. Of the 5463 teeth with intact restorations, 5% had radiographic secondary caries. We found that the likelihood of radiographic secondary caries was nearly three-fold higher for defective restorations than for intact restorations. The large percentage (86%) of defective restorations with no radiographic secondary caries suggests, however, that replacement of all defective restorations due to risk of secondary caries may constitute overtreatment. The use of defective restoration status and presence of radiographic secondary caries as a combined criterion for replacement may potentially reduce such overtreatment. The prevalence of secondary caries under defective restorations should be determined clinically so that the usefulness of combining the criteria of defective restoration with those of radiographic secondary caries can be evaluated.

Introducing dependency into IC yield models

Abstract Many models have been proposed to explain the distribution of point defects on integrated circuit (IC) chips. The most popular of these models are based on distributions derived from the Poisson distribution for describing the number of defects expected on a chip. These models do not inherently allow for dependence between the number of defects expected on two adjacent chips. This paper introduces a model for generating an IC defect distribution that allows for dependence in the number of defects on chips that are near one another. Simulations are done to compare the new dependent model to a previously proposed independent model. These simulations with the new model show defective chips clustered near each other on the water. This clustering property has been observed on real wafers. Methods of parameter and yield estimation in the dependent model are discussed.

Efficacy of FDA guidelines for ordering radiographs for caries detection

This study evaluated the effect on caries diagnosis of using the Food and Drug Administration (FDA) guidelines for ordering radiographs on a sample of new adult patients seeking general dental care at a dental school clinic. These guidelines recommend posterior bitewing views accompanied by supplemental periapical views as indicated by the patients clinical signs or symptoms. We compared the findings from the use of this protocol to those from a full-mouth set of radiographs made on the same new patient. We examined 490 patients and ordered periapical radiographs in accordance with the FDA guidelines, namely, when there was some clinical sign or symptom suggesting the need for radiographs. A full-mouth set of radiographs was obtained, and both the full and selected sets were evaluated separately. We found that 95% of our patients had one or more carious lesions, with an average of 5.7 carious teeth per patient. Of the 2808 carious teeth detected in the study, 1949 (69%) were found radiographically and 1564 (56%) were found clinically. Indeed, 1244 (44%) of the carious teeth were detected only by radiographic examination. Use of the FDA guidelines resulted in a 43% reduction in the number of radiographs ordered. Reduction in the number of images made on patients resulted in failure to detect 93 lesions (3.3%) located on 88 teeth in 11.4% of the study patients. Most of the missed lesions (2.9%) were radiographically confined to the enamel of anterior maxillary teeth. Only 36 of the missed lesions (1.6% of all radiographic lesions) extended into the dentin. We believe that this number is very small and most likely inconsequential compared with the considerable diversity among dentists regarding radiographic detection of caries, as well as the diversity of opinion regarding the indications for restorative treatment.

Confidence intervals for the interrater agreement measure kappa

The asympotic normal approximation to the distribution of the estimated measure [kcirc] for evaluating agreement between two raters has been shown to perform poorly for small sample sizes when the true kappa is nonzero. This paper examines the use of skewness corrections and transformations of [kcirc] on the attained confidence levels. Small sample simulations demonstrate the improvement in the agreement between the desired and actual levels of confidence intervals and hypothesis tests that incorporate these corrections.

Using simulated envelopes in the evaluation of normal probability plots of regression residuals

We investigate properties of a diagnostic-envelope method for evaluating normal probability plots of regression residuals that was proposed by Atkinson (1981), implemented by BMDP (Hardwick 1987), and extended to logistic regression by Landwehr, Pregibon, and Shoemaker (1984). The envelopes stability properties and joint residual vector-inclusion levels, undocumented so far in the literature, are explored here with several examples. Alternative resistant techniques for creating envelopes are considered; interpretations that can be derived from these plots are discussed. A resistant version of the envelopes shows good stability and good sensitivity to outlying residuals; both full-normal and half-normal probability plots with this envelope method provide useful information to the analyst.

Efficacy of the FDA Selection Criteria for Radiographic Assessment of the Periodontium

The diagnosis of periodontitis is generally made on the basis of a clinical examination supported by radiographic evidence of bone loss. Recent guidelines promulgated by the US Food and Drug Administration recommend that periapical radiographs be ordered on the basis of clinical signs and symptoms indicating the probable presence of disease. This study evaluated the effectiveness of the FDA Guidelines for ordering radiographs for new adult dental patients as related to assessment of the periodontal condition of the patient. We examined 490 patients and determined the periapicals needed to supplement the posterior bitewings based upon the patients clinical findings. We measured the reduction in the number of radiographs ordered as well as the extent of missed alveolar and furcation bone loss resulting from the use of the selected set of radiographs compared with a complete set. Four hundred thirty-three subjects had at least one clinical sign of periodontitis present in their mouths, and 264 demonstrated radiographic evidence of alveolar bone loss. Of the 460 subjects on whom periodontal probing was conducted, two-thirds demonstrated periodontal probing depths in excess of 3 mm; almost half showed evidence of bleeding upon probing. Individuals with clinical signs of periodontitis had, on average, 10 periapicals ordered-more than twice the number as those with no sign of periodontitis. Of the 2,415 teeth with radiographic findings of proximal or furcal bone loss, 152 sites of bone loss (6%) were missed when the selected set of films plus the posterior bitewings was used.

Efficacy of FDA guidelines for prescribing radiographs to detect dental and intraosseous conditions

OBJECTIVES This study compared the FDA guidelines for ordering dental radiographs to a conventional full-mouth examination for the detection of intraosseous disease and conditions affecting teeth other than caries. STUDY DESIGN We examined 490 patients and selected posterior bite-wing and periapical views as indicated by the patients signs or symptoms. We compared the radiographic findings with the use of this selected set of radiographs to those from a full-mouth set of radiographs to determine the rates of missed disease when the FDA guidelines were used. RESULTS The most commonly missed intraosseous findings were osteosclerosis, unerupted teeth, periapical radiolucencies, and primary root tips. Periapical radiolucencies that were most probably periapical cemental dysplasia were missed in six patients. The most commonly missed dental findings were resorbed roots and pulp stones. Three instances of dens in dente were missed. CONCLUSIONS When we used the FDA guidelines, the number of missed intraosseous and dental conditions was small and most likely inconsequential given the range of variability in dental diagnosis and treatment.

The analysis of incomplete data in the three-period two-treatment cross-over design for clinical trials

The additional time to complete a three-period two-treatment (3P2T) cross-over trial may cause a greater number of patient dropouts than with a two-period trial. This paper develops maximum likelihood (ML), single imputation and multiple imputation missing data analysis methods for the 3P2T cross-over designs. We use a simulation study to compare and contrast these methods with one another and with the benchmark method of missing data analysis for cross-over trials, the complete case (CC) method. Data patterns examined include those where the missingness differs between the drug types and depends on the unobserved data. Depending on the missing data mechanism and the rate of missingness of the data, one can realize substantial improvements in information recovery by using data from the partially completed patients. We recommend these approaches for the 3P2T cross-over designs.