Robert B. Bendel

Comparison of Stopping Rules in Forward “Stepwise” Regression

Abstract This paper uses the unconditional mean square error of prediction as a criterion for comparing stopping rules used with the forward “stepwise” selection procedure in multivariate normal samples, based on simulations of 48 population correlation matrices. The CP statistic, “F to enter” (.15 < α < .25), a rule which minimizes the sample criterion, and one which sequentially tests the equality of the population criterion (.25 < α < .35) are superior. For these rules, the criterion seldom differs by more than three percent, although there are considerable differences between these and some of the other rules.

Disparities in Smoking Cessation Between African Americans and Whites: 1990-2000

OBJECTIVES We examined disparities in smoking cessation rates between African Americans and Whites from 1990 through 2000. METHODS We performed an analysis of smoking cessation with data from the National Health Interview Surveys of 30660 African Americans and 209828 Whites, 18 to 64 years old, with adjustment for covariates in multiple logistic regression models. RESULTS Whites were significantly more likely than African Americans to be former smokers, and this disparity in the quit ratio persisted from 1990 through 2000. After adjustment for covariates, disparities were substantially reduced especially among women. Among former smokers, African Americans were significantly more likely than Whites to have quit successfully within the past 10 years. CONCLUSIONS Statistical adjustment for covariates reduces African American-White disparities in quit ratios, and recent cessation patterns suggest possible future reductions in disparities.

Fetal erythrocyte phospholipid polyunsaturated fatty acids are altered in pregnancy complicated with gestational diabetes mellitus

Insulin resistance and altered maternal metabolism in gestational diabetes mellitus (GDM) may impair fetal arachidonic acid (AA) and docosahexaenoic acid (DHA) status. The objectives were to test the hypothesis that fetal polyunsaturated fatty acids would be altered with GDM and identify factors related to fetal phospholipid (PL) AA and DHA. Maternal and cord vein erythrocyte PL fatty acids were determined in GDM (n=13) and healthy pregnant women (controls, n=12). Cord vein erythrocyte PL AA and DHA concentrations were significantly lower in GDM vs. controls. Maternal blood hemoglobin A1C was inversely correlated to fetal erythrocyte PL DHA and AA in controls and GDM (n=25). Pregravid body mass index was negatively associated with fetal PL DHA. The data support the hypothesis that there is impairment in fetal accretion of DHA and AA in GDM.

Opinions among mandated reporters toward child maltreatment reporting policies

OBJECTIVE This study investigates the opinions of social workers, pediatricians, and physician assistants in their support for the existing child maltreatment mandated reporting policy and an alternative reporting policy. Differences in professional training, age, gender, reporting behavior, attitudes towards reporting, and level of certainty needed to report were considered in determining which group of mandated reporters was most likely to support either the existing or an alternative policy. METHOD A total of 382 social workers, pediatricians, and physician assistants in Connecticut and Massachusetts completed a self-report questionnaire. This sample represented an overall adjusted response rate of 76%. RESULTS There was slightly more support for the existing reporting policy compared to the alternative policy. After controlling for age, attitudinal variables, and reporting behavior, social workers were significantly more likely to favor an alternative child maltreatment reporting policy than pediatricians or physician assistants. Also, those mandated reporters who had not consistently reported all suspected cases of maltreatment were significantly more likely to favor the alternative policy than consistent reporters. CONCLUSIONS Because the majority of mandated reporters do not consistently report all suspected cases of maltreatment and support for an alternative reporting policy varies among different groups, consideration of alternative strategies and policies that address the concerns of different types of professionals may be appropriate.

Cigarette Smoking Among Native and Foreign-Born African Americans

PURPOSE To examine differences in current smoking status and the number of cigarettes consumed daily between foreign and native-born African Americans, and the impact of demographic and socioeconomic status (SES) factors on smoking behavior. METHODS Data were obtained from combining the 1990-1994 National Health Interview Surveys and consisted of 16,738 U.S. born and foreign-born African Americans between 18 and 64 years of age. The statistical analysis included cross-tabulations and weighted multiple logistic regression (MLR) using the Statistical Packages for the Social Sciences (SPSS) and the Survey Data Analysis (SUDAAN) computer programs. RESULTS Adjusted MLR analysis revealed that native-born African Americans were more likely (odds ratio (OR) = 2.7, p < 0.001) to be current smokers than foreign-born blacks. Within the native-born group, smoking prevalence decreased with increasing education and income, but these associations were not found for foreign-born blacks. Women in both groups were less likely than men to be current smokers. Statistically significant differences were not found between the two groups in the number of cigarettes smoked per day. CONCLUSIONS This analysis of nativity and smoking behavior further demonstrates the social diversity among African Americans and suggests the differential impact of social and cultural factors on smoking behavior within racially classified social groups. In areas where there are substantial numbers of foreign-born blacks, researchers should consider differentiating smoking status by nativity. Though differences in smoking prevalence were apparent for native and foreign-born American Americans, prevention and cessation programs are needed for both groups.

The Effects of Sociodemographic Variables, Training, and Attitudes on the Lifetime Reporting Practices of Mandated Reporters

This study examined the effects of sociodemographic, training, and attitudinal variables on the child maltreatment lifetime reporting proportion (LRP) of 382 randomly selected pediatricians, masters level social workers (MSWs), and physician assistants (PAs). Findings indicated a lifetime mean of 24 suspected cases of child maltreatment and an average of 16 reported cases with an LRP of 69%. Pediatricians reported the highest LRP (76%), followed by MSWs (63%) and PAs (60%); differences in LRP were partially reduced to nonsignificance after adjusting for the other independent variables. Multivariate analysis revealed that the strongest predictors of LRP were case-related attitudes, professional concerns, institutional setting, and the amount of training the professional received. These results demonstrate the importance of professional education and opinions of mandated reporters in reporting practices.

Elevated lipoprotein lipids and gestational hormones in women with diet-treated gestational diabetes mellitus compared to healthy pregnant controls.

The objective of this study was to describe plasma and lipoprotein perturbations in gestational diabetes mellitus (GDM) compared to controls, and determine if alterations in lipids are related to gestational hormones and/or glucose control. Maternal HbA1c, free fatty acids (FFA), beta-estradiol, progesterone, prolactin, and plasma, very-low-density lipoprotein (VLDL), low-density lipoprotein (LDL), high-density lipoprotein (HDL), HDL2 and HDL3 triglyceride (TG), cholesterol, and dietary intake were determined for women with diet-treated GDM and controls in a longitudinal design. Subjects (N = 25/group) were matched for age, race, and body-mass index (BMI). Women with GDM had significantly higher HbA1c than controls, although both groups were within the normal range (4%-6%). However, body weight gain was less for women with GDM. There was a trend for higher plasma FFAs at 37-38 weeks in GDM versus controls. Plasma and lipoprotein TG among the groups increased over the third trimester, and mean concentrations were greater for women with GDM. In GDM versus controls, VLDL and HDL3 TGs were higher at all times, HDL and HDL2 TGs at 33-34 and 37-38 weeks, and LDL TGs at 37-38 weeks. In VLDL, core lipids (TG + cholesterol) increased over gestation and were greater in GDM. In HDL, the TG/cholesterol ratio was greater in GDM. In GDM versus controls, plasma progesterone and prolactin were higher at all times; beta-estradiol was elevated at 37-38 weeks. HbA1c, progesterone, and prolactin correlated with all lipoprotein TG fractions. Exaggerated hypertriglyceridemia, particularly in the VLDL and HDL fractions, is a feature of GDM. The increase in VLDL TG is likely due to an increase in VLDL synthesis, whereas particle enrichment in TG is a plausible explanation for changes in HDL TG. Slight perturbations in glucose control and gestational hormones in diet-treated GDM may contribute to the observed increase in plasma and lipoprotein TG.

Population correlation matrices for sampling experiments

A procedure is given for generating correlation matrices which can be used as population correlation matrices for sampling experiments. The algorithm specifies the eigenvalues and randomly selects a correlation matrix from the class of all correlation matrices which possess these same eigenvalues. It is possible to obtain a set of correlation matrices which are indexed by the degree of interdependence among the variables by parameterizing the eigenvalues with a single parameter. An example is the case in which the eigenvalues form a geometric progression. Examples are given and an application to the problem of stopping rules in stepwise regression is discussed. Other applications are also briefly discussed.

Storytelling Intervention for Patients With Cancer: Part 2—Pilot Testing

PURPOSE/OBJECTIVES To evaluate symptom reports and the impact of a nurse-led storytelling intervention in a supportive group setting on mood, stress level, coping with stress, pain, self-efficacy, and satisfaction with life in patients with cancer. DESIGN Descriptive pilot project using a pretest/post-test control group. SETTING Local regional medical center in the Pacific Northwest region of the United States. SAMPLE Convenience sample of 10 patients with various cancer diagnoses; 7 completed the intervention. METHODS Participants were randomly assigned to a storytelling or control group. Using a tool kit generated for this project, a nurse facilitator guided storytelling group participants in 12 1.5-hour sessions. Six instruments, symptom assessments, and a retrospective physician chart review were completed for each group. Data were analyzed using repeated measures analysis of variance. MAIN RESEARCH VARIABLES Mood, stress, coping, pain, self-efficacy, and satisfaction with life. FINDINGS Comparison of changes in group mean scores revealed a significant decrease in anxiety in the storytelling group despite disease progression. Documentation of psychosocial symptomatology by physicians is limited; however, nursing assessments were useful in determining psychosocial status before and after the intervention. CONCLUSIONS Results can be viewed only in context of a feasibility study and are not generalizable because of a limited sample size. A trained oncology nurse was able to use the storytelling intervention. Initial results are promising and warrant further study. IMPLICATIONS FOR NURSING After additional testing, the intervention could be used to enhance storytelling groups for patients with cancer or for individuals who are uncomfortable in or do not have access to storytelling groups.

Social heterogeneity in smoking among African Americans.

OBJECTIVES This study assessed the effects of sociodemographic variables on the smoking behavior of African Americans. METHODS Data for 14,397 African-American respondents (18-64 years old) to the National Health Interview Survey in the years 1990 through 1993 were examined in multivariate models. RESULTS Age was the strongest predictor of smoking, and men were at least 1.75 times as likely as women to be smokers (P < .001). CONCLUSIONS The Year 2000 goal of an 18% prevalence rate among African-American adults may not be attained without major community interventions focusing on male and middle-aged African Americans.