Subset Regression Analysis Research Articles

Establishment of a disaster-related psychological screening test.

To create a short screening scale for the detection of posttraumatic stress disorder (PTSD) and major depressive disorder (MDE) in earthquake survivors in Taiwan. Trained research assistants used the Disaster-Related Psychological Screening Test (DRPST) to assess 461 residents of a village that had experienced a major earthquake. The participants were also evaluated by psychiatrists using the Mini-International Neuro-psychiatric Interview (MINI). Best subset regression analysis and the receiver operating characteristics curve were used to select a subset of items and cut-off points from the DRPST. A seven-symptom scale and a three-symptom analogue were selected for PTSD and MDE screening, respectively. Scores of three or more on the PTSD scale and two or more on the MDE scale were used to define a group of positive cases that provide useful information for the patient cohort and will be valuable in long-term follow-up studies of the prevalence of psychiatric diseases following a natural disaster. However, higher scores could also be used to define positive cases under limited psychiatric care resources. The DRPST, which was administered for phase 1 of this two-phase study, may be used for effective and rapid screening for PTSD and MDE after an earthquake, despite the usual limitations on resources following a disaster.

Role of antioxidant nutrients and lipid peroxidation in premature infants with respiratory distress syndrome and bronchopulmonary dysplasia.

The objective of this study was to determine if newborn premature infants with severe respiratory distress syndrome (RDS) who developed bronchopulmonary dysplasia (BPD) demonstrate, within the first 3 days of life, lower blood levels of antioxidants and higher urine levels of lipid peroxidation products than premature infants who recovered from RDS. Perinatal variables (gestational age, birth weight, and Apgar scores) and antioxidant indices in cord and in third day of life plasma and red blood cell (RBC) samples from healthy premature infants (n = 35), infants with RDS (n = 23) and infants with BPD (n = 23) were examined. Antioxidant indices included selenium, alpha-tocopherol, total and oxidized glutathione, glutathione peroxidase, superoxide dismutase, and urinary malondialdehyde. By inferential statistics, only the perinatal variables and cord plasma selenium distinguished healthy premature infants from premature infants with RDS or BPD. From perinatal variables and antioxidant indices we calculated: (1) cord to third-day-of-life variable differences, (2) variable-to-variable ratios, and (3) ratios of a difference for one variable to a difference for any second variable. Subset regression analysis yielded an equation (adjusted R2 = 0.8839) that correctly predicted infants who developed BPD 100% of the time. Predictor variables for BPD were gestational age, Apgar at 1 min, cord and third-day-of-life RBC selenium, cord total glutathione, cord and third-day-of-life glutathione peroxidase and nine different ratios involving Apgar scores, RBC selenium, total and oxidized glutathione, alpha-tocopherol, glutathione peroxidase, and superoxide dismutase. In this study, there was no relationship between lipid peroxidation and BPD. There was a higher rate of patent ductus arteriosus, congestive heart failure, and retinopathy of prematurity in infants with BPD. This study confirms that low plasma selenium and alpha-tocopherol levels in premature infants (< or = 30 weeks' gestational age or lower) were significantly associated with an increased respiratory morbidity.

Abundance and biomass of heterotrophic flagellates, and factors controlling their abundance and distribution in sediments of Botany Bay.

The abundance and biomass of heterotrophic flagellates were estimated monthly in sediments of Botany Bay during March 1999-February 2000. The annual abundance and biomass were in the ranges of 0.46-4.70 x 10(5) cells/cm(3) and of 0.30-8.61 micro g C/cm(3), respectively. The majority of heterotrophic flagellates (93-100%) were less than 10 mm in length and few flagellates were larger than 10 mm. Of the total microbial carbon biomass, heterotrophic flagellates made up about 5% (but at times up to 35%). The contribution of heterotrophic flagellates varied from month to month, and among the sites. The abundance of heterotrophic flagellates was negatively correlated with sediment grain size and positively correlated with the abundance of bacteria, algae (autotrophic flagellates and diatoms), and their probable grazers. A best subsets regression analysis showed that bacterial and algal abundance are the most important factors controlling the abundance of heterotrophic flagellates. When the previously reported grazing rates on bacteria were applied, heterotrophic flagellates would consume a maximum of 64% of bacterial standing stock daily in Botany Bay, suggesting that heterotrophic flagellates are important as bacterivores. However, the importance of heterotrophic flagellate grazing probably varies significantly among the sites and from month to month.

Meperidine analgesia and delirium in aged hip fracture patients

Delirium is quite frequent in elderly patients who sustain hip fractures. The use of Meperidine by physicians, unaware of the possible emergence of delirium in elderly patients, is very popular. We have retrospectively examined the incidence of delirium in 181 consecutive patients admitted to the orthogeriatric ward with hip fractures. We used the confusion assessment method to establish the presence of delirium in all patients. A database search was conducted to identify which patients were treated by Meperidine, or Morphine, prior to delirium onset. We identified 92 cases, 44 of whom were treated by Meperidine alone, and the other 48 treated by Morphine alone. Delirium was diagnosed in 13 (27.1%) Morphine treated patients as compared with 19 (43.2%) treated by Meperidine ( P<0.001). Age, cognitive status and opiate use were associated with perioperative delirium. A subset regression analysis showed that exposure to Meperidine was significantly associated with delirium (odds ratio 2.5, P<0.01), in contrast with Morphine. Our results confirm the association between exposure to Meperidine and delirium, suggesting that this drug should be withdrawn in elderly hip fractured patients undergoing surgery, and substituted by low dose Morphine analgesia. Reducing the incidence of delirium, by adopting such an approach, may result in a significant potential of savings in direct costs, related to treatment of delirium in this population.

Regression analysis of experiments with complex confounding patterns guided by the alias matrix

Resolution III experiments with complex confounding patterns, often called main effect plans. Their main effects are partially confounded with two-factor interactions rather than being either independent or completely aliased, as they are in regular designs that are constructed from a defining relation. It is possible to detect some two-factor interactions, from experiments with complex confounding patterns if only a few of the factors are active. The partial confounding of interactions with main effects in these experiments allows one to estimate interactions with regression analysis. Recent papers have shown how the interactions can be detected using repeated runs of stepwise regression, or computationally intensive Bayesian methods. This paper shows that, with a short list of candidate interactions, the important interactions can be detected efficiently by a single pass of all subsets regression. The short list of candidates interactions for the all subsets regression are selected with regard to their potential for contributing to the value of the large estimated effects calculated from the data. The potential is determined by summing coefficients in the alias matrix. This paper reviews the concept of the alias matrix, and defines the alias plot that can be obtained from the alias matrix. The alias plot provides simple graphic for viewing the potential contribution of interactions to large estimated effects. The interactions identified on the alias plot, along with main effects, are used as candidates in an all subsets regression analysis of data from partially confounded experiments. The resulting strategy allows exploration of a wide range of the potential model space, can reveal several plausible models for the data if they exist, and can be completed using standard statistical software. Examples are presented.

A novel predictive model of outcome in de novo AML based on S-phase activity and proliferative response of blast cells to haemopoietic growth factors

This study assesses whether the kinetic response of AML cells to HGFs might help to predict initial clinical outcome of treatment in de novo AML in association with age, FAB type and karyotype. Best subset regression analysis indicated optimal variables to develop models to predict prognosis. High S-phase in surviving cells following 7 days incubation in SFM, resistance to stimulation by G+GM-CSF and poor karyotype taken in combination correctly predicted outcome in 83% of patients. The importance of high SFM S-phase may be to indicate autonomous proliferation therefore a leukemic clone more likely to regenerate following therapy at the expense of normal haemopoiesis. Kinetic studies of AML cells may be a useful predictor of outcome in addition to other more established prognostic factors.

PLANT GROWTH AND SOIL METAL CONCENTRATIONS A SPATIAL EFFECTS MODEL

In situ reclamation techniques are proposed for remediation of areas affected by metal mine wastes in Montana’s Clark Fork River Basin. In 1990, an in situ reclamation project was implemented on a fluvially-deposited tailing areas adjacent to the Clark Fork River. As part of this project, a 2.3 ha field containing tailing deposits was tilled to 1.2 m, lime was incorporated into the soil, and rangeland grasses were initially planted. In June of 2000, the field was plowed and seeded with six-row barley (Hordeum vulgare). Variable growth of this species was observed. It was hypothesized that plant growth was negatively correlated with metal and arsenic concentrations in the soil. Variability in barley growth attributable to other measurable soil characteristics was statistically quantified and modeled to account for the effects of landscape spatial heterogeneity. A stratified sampling method was employed for plot selection, with plots selected based on plant height. Three short, five medium, and four tall plant plots were selected. Within each plot, two vegetation and co-located soil samples (0-15 cm) were collected. The <2 mm fraction of the each soil was analyzed for factors related to plant growth and total elemental concentrations of: arsenic (As), cadmium (Cd), copper (Cu), lead (Pb), and zinc (Zn). Elemental data were first reduced using Principal Components Analysis (PCA) to provide a primary “metals” predictor variable for subsequent regressions. All subset regression analyses were used on the principal components to determine which component was the best predictor of plant biomass. Spatial regression analyses were then performed to assess whether a model that accounts for spatial heterogeneity in the landscape was necessary, or whether ordinary multiple regression techniques adequately modeled plant response to soil metal concentrations adequately. Metal concentrations in the soil were the only statistically significant predictors of plant biomass among all factors tested. No spatial autocorrelation was found in the residuals of the ordinary least squares model used. Therefore, a spatial regression model was not required to explain the relationships. Additional

Experimental Validation of a Structure–Activity Relationship Model of Skin Irritation by Esters

SAR model development should be a continuous process involving formulation then experimental testing of the model, incorporation of test results into the database, then refinement of the model using the expanded database. The larger database affords greater confidence in its ability to predict the biological response. This iterative procedure was employed with a recently developed structure-activity relationship (SAR) model of human skin irritation. Based on a “leave-one-out” cross validation, the mean sensitivity of the initial model was 0.89, the mean specificity was 0.74. A clinical validation study was conducted to assess the ability of the model to predict human skin irritation by esters commonly used as fragrance ingredients. Esters that were found to cause irritation in rabbits, and that were within the predictive space of the SAR model, were selected for human testing using the patch test procedure. Of the 34 rabbit irritants selected, 16 were predicted by the model to be positive and 18 were predicted to be negative. Patch testing yielded two positive esters, allyl heptanoate and allyl cyclohexanepropionate. These test results were incorporated into the database to refine the SAR model. Best subsets regression and linear discriminant analysis were used to generate 10 submodels consisting of 10 irritants and 50 non-irritants randomly selected from the new database. Physicochemical parameters associated with irritant esters, when compared with non-irritant esters, differed somewhat from those identified in the original model. Irritant esters had lower solubility parameter and water solubility, higher Hansen dispersion and Hansen hydrogen bonding, and lower sum of partial positive charges, when compared with non-irritant esters. The sensitivity of the new model is 0.69 and specificity is 0.67. The results of this study indicate that SAR models based on limited data may not accurately predict the activity of unknown chemicals even though the computationally-derived sensitivity and specificity of the models are high. This finding emphasizes the need for experimental validation of models and their refinement as new data become available.

Cognitive, Affective, Personality, and Demographic Predictors of Foreign-Language Achievement

The ability of cognitive, affective, personality, and demographic variables to predict second-language acquisition among college students was investigated. An all possible subsets regression analysis was used to compare the proportion of variance in foreign-language achievement explained by each variable. The analysis revealed that variables from each of the 4 domains were important predictors of foreign-language achievement. Overall, academic achievement, as measured by GPA average, was the best predictor, explaining 11.5% of the variance in foreign-language achievement. Foreign-language anxiety, the next best predictor, explained 10.5% of the variance. The educational implications of these findings for understanding foreign-language achievement are discussed, along with suggestions for future research.

A robust structure-activity relationship (SAR) model for esters that cause skin irritation in humans.

A structure-activity relationship (SAR) model has been developed to discriminate skin irritant from nonirritant esters. The model is based on the physicochemical properties of 42 esters that were tested in humans for skin irritation. Nineteen physicochemical parameters that represent transport, electronic, and steric properties were calculated for each chemical. Best subsets regression analysis indicated candidate models for further analysis. Regression analyses identified significant models (p < 0.05) that had variables that were also significant (p < 0.05). These candidate models were evaluated using linear discriminant analysis to determine if the irritant esters could be discriminated from nonirritant esters. The stability of the model was evident from the consistency of parameters among ten submodels generated using multiple random sampling of the database. The sensitivity of the ten models, evaluated by "leave-one-out" cross-validation, ranged from 0. 846 to 0.923, with a mean of 0.885 +/- 0.025 (95% CI). The specificity ranged from 0.615 to 0.923, with a mean of 0.738 +/- 0.06 (CI). Compared with nonirritant esters, irritant esters had lower density, lower water solubility, lower sum of partial positive charges, higher Hansen hydrogen bonding parameter, and higher Hansen dispersion parameter. The results indicate that physicochemical features of esters contribute to their ability to cause skin irritation in humans, and that chemical partitioning into the epidermis and intermolecular reactions are likely important components of the response. This model is applicable for prediction of human irritation of esters yet untested.

Plasma acylation stimulating protein, adipsin and lipids in non‐obese and obese populations

Acylation stimulating protein (ASP) is a potent stimulator of TG synthesis in human adipocytes. In the present study, we have analysed plasma ASP and adipsin levels and their relationships to plasma lipids in non-obese and obese groups. The results show that the frequency distribution of ASP is skewed but that of adipsin is normal in both groups. In the non-obese population, the mean levels of plasma ASP and adipsin were 20.2 nmol L-1 (median) and 66.6 +/- 19 nmol L-1 (mean) respectively. No difference was observed between men and women for each of the parameters. In the obese population, the median plasma ASP was increased by 246% (69.9 nmol L-1) and adipsin by 31% (87.0 +/- 22.7 nmol L-1) above that of the control group. Although the levels for men and women were not statistically different for adipsin, the median ASP plasma concentration was 1.9-fold higher in obese women than in obese men (71.8 nmol L-1 vs. 37.6 nmol L-1, P < 0.05). Best subset regression analysis provided a model with variables that best predict plasma ASP [r2 = 0.160, P < 0.008 for body mass index (BMI), P < 0.05 for triacylglycerol (TG), P < 0.03 for free fatty acid (FFA)] and plasma adipsin (r2 = 0.057, P < 0.017 for BMI) in a non-obese population. In obese subjects, the model was different for plasma ASP (P = NS for any of the variables) and plasma adipsin (r2 = 0.356, P < 0.008 for FFA, P < 0.0002 for BMI, P < 0.02 for age). There was no correlation between ASP and adipsin in either the non-obese or the obese group. The present data suggest involvement of the ASP/adipsin pathway in the pathogenesis of obesity.

Family coping and bereavement outcome.

Using a three-phase longitudinal design, the bereavement of 115 adult Australian families following the death of a parent from cancer was studied. The cohort comprised 115 spouses and 153 offspring: 670 individual responses were obtained. A range of psychosocial variables was studied through a semistructured interview covering the experience of the deceased's illness, death and funeral, spousal health, family coping, sources of support, use of ritual and completion of estate duties. Bereavement outcome was measured by standardized questionnaires of the intensity of grief (Bereavement Phenomenology Questionnaire), depression (Beck Depression Inventory), psychological distress (Brief Symptom Inventory) and social adjustment (Social Adjustment Scale). Those psychosocial variables found to be significantly correlated with bereavement outcome were entered into best sub-set regression analyses. Family coping was the most consistent correlate of bereavement outcome in these regression analyses, which accounted for up to 38% of the variance in grief, 64% in distress, 53% in depression and 46% in social adjustment. The nature of family functioning is a key aspect of social support in influencing the outcome of bereavement.

Short forms to assess life quality and symptom distress for urinary incontinence in women: The incontinence impact questionnaire and the urogenital distress inventory

This article describes short form versions of the Incontinence Impact Questionnaire (IIQ) and the Urogenital Distress Inventory (UDI). These instruments assess life impact and symptom distress, respectively, of urinary incontinence and related conditions for women. All subsets regression analysis was used to find item subsets that best approximated scores of the long form versions. The approach succeeded in reducing the 30-item IIQ and the 19-item UDI to 7- and 6-item short forms, respectively. The short form versions may be more useful than the long form versions in many clinical and research applications.

Comparison of protein extraction procedures to assess the effect of quantitative protein composition on the bread-making quality of South African winter wheat cultivars

A suitable technique for the extraction of wheat flour protein fractions was pursued to assess the effect of quantitative protein composition on the bread-making quality of winter wheat cultivars on a large scale. Nine procedures involving the use of different sequential solvent extractions were quantitatively compared. The distribution of protein components in the different solubility classes was checked by means of sodium dodecyl sulphate Polyacrylamide gel electrophoresis. The most effective procedure to fractionate the four major protein groups sequentially was: albumin (water),' globulin (0.5 M NaCI + 0.05 mM Na-EDTA), gliadin [55% (v/v) propan-2-ol] and glutenin [0.0125 M Na2B4O7, 0.043 M NaOH, 0.5% (m/v) SDS, 1% (v/v) β-mercaptoethanol].To validate this procedure, linear correlations between quality parameters and protein extracted for all localities and cultivars were complemented by principal component and best subset regression analysis. Albumin and globulin were weakly correlated with quality a...

Nutrition and Serum Cholesterol Levels Among Elderly Men and Women (Dutch Nutrition Surveillance System)

Associations of serum cholesterol with relevant dietary intake variables (assessed with the dietary history method) and body mass index were investigated in elderly men (n = 199) and women (n = 180) 65-79 years old. All subjects were apparently healthy, nondiabetic, and not on a dietary regimen. The associations were studied separately for men and women using linear regression analysis and all possible subsets regression analysis. Among men, body mass index (kg/m2) and intake of monounsaturated fat and of alcohol were positively and consistently associated with serum total cholesterol. Among women, intake of alcohol and of saturated fat were positively associated, and intake of polysaccharides was inversely associated with serum total cholesterol. The intake of monounsaturated fatty acids was highly (r greater than .60) positively correlated with the intake of total fat and saturated fatty acids, and inversely with carbohydrates. HDL-cholesterol was positively associated with alcohol intake (significant for men only), and inversely with body mass index (women). The results indicate that the effect of dietary factors on serum cholesterol levels is probably not age-limited. Elderly people may potentially benefit from weight reduction or control, moderate alcohol consumption, and avoidance of too much dietary fat. These suggestions are in fair accordance with general population-based guidelines for a healthy diet. However, as our study was cross-sectional, causation as well as the public health impact remains to be proven.

The ecology of Ramalina menziesii. V. Estimation of gross carbon gain and thallus hydration source from diurnal measurements and climatic data

All subsets regression analysis was used to determine the set of meteorological variables that best predicted daily gross carbon fixation of the lichen Ramalina menziesii Tayl. measured on 23 days at Hastings Reservation in central coastal California. The regression equation containing the two best predictor variables, the duration of diurnal atmospheric humidities equal to or above 90% and daily precipitation, was then used in conjunction with weather records to estimate daily, seasonal, and annual gross CO2 gain of R. menziesii at Hastings for the years 1981 – 1983. Calculated average annual gross CO2 fixation for these 3 years was 2756 mg CO2∙g dry weight−1. For 66% of this total gain, the most likely source of thallus hydration was rainfall. Dewfall was responsible for 15%, atmospheric water vapor uptake for 11%, and for the remaining 8% the source was uncertain. Three-fourths of the annual carbon fixation took place during the 6 months of the rainy season.